Full text of "Paradigms Lost. Tackling the Unanswered Mysteries of Modern Science"

See other formats

HBBHH

"Provocative and delightful... An overview of some of the most
tantalizing unsolved problems facing scientists today" Robert Shapiro

TJl C A "V C7 L U_

l/lviilcA-LCJ j Mod.C’VW ^CICAACC

5v\j,ervc<

JOHN L. CAST!

I j

TRADE PAPERBACKS
U.S. $15.95

Can. $24.50

Avon Books are available at special quantity discounts for bulk
purchases for sales promotions, premiums, fund raising or educa-
tional use. Special books, or book excerpts, can also be created to
fit specific needs.

For details write or telephone the office of the Director of Special
Markets, Avon Books, Dept. FP, 1350 Avenue of the Americas,
New York, New York 10019, 1-800-238-0658.

Oil ™ iL

uslcvlcs of Mc^clcv^

/CV.CWCC

JOHN L. CAST!

AN AVON BOOK

Permissions on page 567

AVON BOOKS
A division of
The Hearst Corporation
1350 Avenue of the Americas
New York, New York 10019

Cover art copyright © 1990 Escher Heirs/Cordon Art-Baam, Holland
Published by arrangement with William Morrow and Company, Inc.

Visit our website at http://www.AvonBooks.com/Bard
Library of Congress Catalog Card Number: 88-37232
ISBN: 0-380-71165-6

All rights reserved, which includes the right to reproduce this book or portions thereof in
any form whatsoever except as provided by the U.S. Copyright Law. For information
address Permissions Department, William Morrow and Company, Inc., 1350 Avenue of
the Americas, New York, New York 10016.

The William Morrow and Company edition contains the following Library of Congress
Cataloging in Publication Data:

Casti, J.L.

Paradigms lost : images of man in the mirror of science / John L. Casti.

p. cm.

Includes index.

I. Science — Miscellanea. I. Title
Q173.C35 1989 500— dcl9

88-37232

CIP

Bard Printing: March 1998

First Avon Books Trade Printing: November 1990

BARD TRADEMARK REG. U.S. PAT. OFF. AND IN OTHER COUNTRIES, MARCA REGISTRADA, HECHO
EN U.S. A.

Printed in the U.S.A.

OPM 10 9 8

If you purchased this book without a cover, you should be aware that
this book is stolen property. It was reported as “unsold and destroyed”
to the publisher, and neither the author nor the publisher has received
any payment for this “stripped book.”

To anyone who’s ever wondered “Why?”
and especially to those visionaries whose work described here
takes us well down the road to “Because!”

PREFACE

Quo Vadimus? The eternal question: “Where are we going?” In
more colloquial terms, we might ask, “Where are you coming
from?” And if we’re in a contemplative mood, we might even
vary the basic theme and extend our inquiry to arrive at that
deepest question of all speculative thought: “What is the true
nature of mankind?” Fundamentally, this is a book putting
forth science’s best guesses regarding the ways to assemble the
pieces of this ultimate, eternally difficult, and ever-tantalizing
puzzle. More precisely, each of the scientific stories I’ve chosen
to tell here addresses in its own characteristic way the issue of
the uniqueness of mankind, in our lives here on Earth, our place
in the galaxy, and even our role in the universe at large. In
short, our most basic concern here is to explore what science has
to say about the perpetually elusive question “Is there anything
special — or unique — about human beings?”

Eternal questions have a nasty habit of remaining eternally
impenetrable when left on the lofty plane of philosophical dis-
course. Consequently, I’ve tried to decompose the “uniqueness of
mankind” question into a set of bite-sized, individually digest-
ible pieces involving our human (1) physical and biochemical
structure, (2) social behavioral patterns, (3) linguistic communi-
cation capabilities, (4) cognitive thought processes, (5) presence
in the galaxy, (6) role as observers in the universe. Each of these
aspects of our lives and activities is paired with what I think of
as one of the Great Problems of modern science: the origin of
life, sociobiology, language acquisition, thinking machines, the
search for extraterrestrial intelligence, and quantum reality. As
Francis Crick once remarked in a similar context, “To show no
interest in these topics is to be truly uneducated,” a good exam-
ple of the well-known Crick irony. Personally, I would vary
Crick’s thesis a bit by saying that to show no interest in these
topics is to be uninformed about the true nature and beauty of
the problems. My hope is that by telling the story of where sci-
ence stands today on each of these problems, it will be possible to
shed light on the more general question of where Homo sapiens

PREFACE

viii

as a species fits into the cosmic scheme of things. I would also
hope to make a small contribution to education by displaying a
few of the fascinating interconnections between these seemingly
diverse pebbles strewn about on the seashore of science.

One of the more deceptive aspects of a scientific research arti-
cle is that the path of development from hypothesis to conclusion
as traced in the paper is almost never a faithful account of how
the results were really obtained. So it is with this book as well.
The noble theme of our special position in the cosmos never en-
tered my mind when I set forth on this project. My original
aims were far more modest, involving little more than trying to
trace for myself and my students the multiple threads of a num-
ber of interesting questions scattered across the landscape of
modern intellectual thought. It was only after I began to gather
together these individual strands that it dawned on me that the
work’s overall theme is really what I’ve grandly termed the
uniqueness of Mankind.

Originally this book came about as the outgrowth of another
work of mine, Alternate Realities: Mathematical Models of Nature
and Man (New York: Wiley, 1989), which is a semitechnical
textbook on the modeling of natural and human systems. While
preparing that volume, I had occasion to wander over a pretty
diverse mindscape of topical areas ranging from chaos, game
theory, catastrophe theory, and cellular automata to their ap-
plications in physics, engineering, evolutionary biology, and cog-
nitive psychology, as well as ecological and economic cycles and
beyond. More than ever before, this venture brought home to me
the heretofore unappreciated fact that virtually all knowledge is
intertwined at some level in nontrivial ways, and that these in-
terrelationships are important enough in their own right that
they should be included in the course programs of all serious
students and aspiring researchers. This conviction led to my re-
solve to put together a course of lectures for the general science
student. These lectures focused on several topical areas in mod-
ern intellectual thought in which the central problems lie in that
no-man’s-land between the boundaries of the classical disci-
plines. It is these Great Problems — the origin of life, sociobiol-
ogy, quantum reality, and all the rest — that form the heart of
this book. As a result of the overwhelmingly encouraging re-
sponse to the lectures, I felt emboldened enough to undertake
the task of trying to put the competing ideas, approaches, and
personalities down on paper in a form accessible not just to uni-

PREFACE

versity students, but to the proverbial educated layman as well.
The result is the book you now hold in your hands.

When reflecting on the volume’s overall structure, I’m contin-
ually reminded of one of those frosted layer cakes that certain
Viennese pastry shops of my acquaintance specialize in serving
up to their gluttonous clientele. The deepest layer running
through every chapter of the book is the eternal question dis-
cussed above: How special are humans, either here on Earth or
in the universe? Set on top of this delectable base is a second
layer consisting of the individual stories taken on their own mer-
its: How does science come to its conclusions about what is
“true”? How did life get its start here on Earth? Are our social
behavioral patterns “programmed” into our genetic makeup?
What is the mechanism by which we learn to speak? Can ma-
chines think? Are there other intelligent beings in the Milky
Way? Does reality itself require our presence as observer/par-
ticipants? Even without the underlying philosophical layer, the
individual stories are sweet and juicy enough in their own right
to provide anyone with a tasty, intellectually fattening treat. Fi-
nally we have the frosting on the cake: the scientists themselves
seen in all their glory (and some of their frailty, too) as they act
to close the circle of self -reference in their singular role as hu-
mans investigating humanity. Taken as a whole, this particular
layer cake is, I think, one that any of my favorite Konditoreien
would proudly feature on its menu of irresistible attractions.

For the sake of exposition, I use the format of a jury trial to
present the competing positions on each of the topical issues of
the book. Consistent with this courtroom motif, each chapter be-
gins with a “claim” phrased to represent the Prosecution’s
charge. The negation of that basic claim constitutes the position
of the Defense. Following the customary sequence for jury tri-
als, each chapter proceeds through opening statements, witnesses
for the Prosecution and Defense, testimony from expert wit-
nesses, summary arguments, and finally the verdict. In this last
connection, I step out of my role as court reporter, don the hat
of a typical member of the jury, and try to assess the merits of
the competing arguments from the position of an uninvolved,
but keenly interested, neutral observer. It’s my hope and expec-
tation that each reader will also serve as a member of this jury,
coming to his or her own conclusions at the end of the competing
arguments.

In attempting to address such a wide array of topics within

PREFACE

the confines of a few hundred pages, compromises necessarily
had to be made. On the one side, to do justice to the ideas, argu-
ments, and genius of the various scholars, I have perhaps de-
scribed some of the material in a bit more detail than the
average reader might care to confront head on. But if you find
yourself starting to lose sight of the forest for the trees, don’t
despair. To help you stay in the game, I have employed several
types of attention-focusing devices. First of all, in each chapter
where the sheer weight of terminology begins to become burden-
some, I have inserted a terminological table at a strategic loca-
tion early in the proceedings. This table can be used as a
convenient point of reference for the nomenclature as you wend
your way through the arguments that follow. But the argu-
ments themselves are not of uniform difficulty, either in the con-
cepts they expose or in the twists and turns of their logic.
Consequently, I have provided a variety of amplifying remarks
in the notes and references for each chapter, material to which
the reader can turn for further elaboration of some of the trick-
ier-than-average passages of the chapter itself. Finally, each
chapter is liberally sprinkled with a number of figures and dia-
grams that I trust will illustrate the main points far more effi-
ciently and clearly than any amount of prose ever could. It is
hoped the combination of these various devices will enable the
general reader to stay afloat while navigating through the more
dangerous rapids of our fast-flowing stream of knowledge.

At the other end of my potential reader spectrum are profes-
sional researchers and students. To these experts I offer my
sympathies for what must appear at times to be gross carica-
tures of their beloved disciplines. My only defense is that such
an approach is necessary in a broad, general treatment of this
sort. As partial compensation, I trust that the admittedly incom-
plete treatment of the expert’s territory given here will at least
bring that territory to the attention of a wider audience, thus
focusing a few rays of the public spotlight where it might do
some good. Finally, I should recall here the fact that the book
originally arose out of a course of lectures for both university
students and faculty. These lectures were somewhat more techni-
cal and detailed than the treatment in the book, containing far
more material from the professional literature, more mathemati-
cal pyrotechnics, more finely detailed arguments, and so forth.
For those readers who want to examine this additional material

PREFACE

or, perhaps, use this book as the basis for a lecture course of
their own, I will be happy to provide my raw research notes,
containing many additional references and sidelights that for
various reasons didn’t find their way into the book itself. Read-
ers wishing to obtain this material should contact the author c/o
Institute for Econometrics, OR, and System Theory, Technical
University of Vienna, Argentinierstrasse 8, A-1040 Vienna,
Austria.

So on balance, in attempting to navigate the fine line between
boring the professional and overwhelming the layman, I tried to
follow here what I think of as the Three-A”s-Minus-One Rule:
make the book educational, enlightening and entertaining with-
out making it encyclopedic. As Anatole France once remarked,
“I prefer the errors of enthusiasm to the indifference of wis-
dom.” But as always, I’ll let the reader be the final judge of the
degree to which I have succeeded in walking this tightrope be-
tween triviality and impossibility.

A quick peek at the Contents will probably generate the im-
pression that each of the book’s chapters is an independent mod-
ule that can be read without reference to any of the others. To
confirm your suspicions, this is indeed the case. I had two con-
siderations in mind when structuring the book in this way. The
first was to reshuffle the deck every now and then, so that if you
run hopelessly aground somewhere along the line, salvation is no
more than a few pages away. And second, while both Francis
Crick and I would find it hard to understand how anyone could
fail to be interested in every one of the topics dealt with here,
empirical observation forces me to the unhappy conclusion that
this really could happen — such people do indeed exist! So if your
tastes run toward extraterrestrial life and you couldn’t care less
about the genetic determination of human behavior, you may
with confidence proceed directly to Chapter Six. Or if you’re
worried about a thinking machine’s taking over your job (or
your life), you may safely skip our deliberations on the origin of
life and move with all due dispatch to Chapter Five. Without
exception, each chapter is totally independent of the others, and
you won’t be hindered in the slightest if you just open the book
at random and start reading.

But while I’m dispensing this largesse, let me introduce a
precautionary note as well. If you want to get right down to the
fundamental layer of the cake constituting the entire book and

XII

PREFACE

learn about the uniqueness of mankind, then the more chapters
you read, the better position you’ll be in to understand the many
facets of the problem and the truly staggering magnitude of the
task involved in providing even a partial answer. Consequently,
if it’s the essence of humanity you’re after, my recommendation
is at least to skim all the chapters. Some of them, like the chap-
ter on thinking machines or the one on life’s origins, involve
slightly more abstract notions and hence are probably a bit
tougher sledding than the average reader might want to enter
into immediately. Nevertheless, each chapter is a piece in the mo-
saic of mankind, and to see the Big Picture you need to know at
least something about the Great Problems — all of them. So skim
if you must, but do so at your own peril.

A last bit of advice on reading the book: Don’t start with
Chapter One! I’m sure this admonition would strike my old
high-school English teacher as nothing short of sheer heresy. Nev-
ertheless, there is at least a little method in this seeming mad-
ness. I have chosen the ordering of the book’s chapters to reflect
a certain progression from life to behavior to mind, from Earth
to galaxy and beyond. The opening chapter is designed to pro-
vide the philosophical and sociological underpinnings to the
scientific doings recounted in following this progression. So
from a logical standpoint, the chapter ordering is airtight and
almost foreordained. However, experience shows that most peo-
ple are like me when they get a new toy (or computer program):
They want to start playing with it right away. And the last
thing they want to do is read the instruction manual cover to
cover before they begin having some fun. So think of the first
chapter as constituting the book’s instruction manual. But since
we all know that you can have lots of fun without knowing the
rules (or, at least, without following them), my advice is initially
to pick out one or two of the topical chapters that capture your
fancy. After digesting this material and getting a feel for how
science operates in practice, you can then go back and compare
how things really work with the way theory and armchair specu-
lation say they should.

Several months ago during the course of discussing this pro-
ject with a colleague, I made the offhand remark that I certainly
hoped that the book would turn out to be a success. Unfortu-
nately, he isn’t the type of friend to let me get away with any
such throwaway remark. “So what is your personal criterion for

PREFACE

xiii

success?” he asked. Resisting the natural impulse to say sales of
a hundred thousand copies (or more) on day one, together with
glowing reviews in all the right places, I finally replied that I
would consider the whole effort to have been worthwhile if I sat
next to someone on a long flight who was reading the book, and
at the end of the flight this nameless companion turned to me
and asked, “Have you read this book?” At this moment, disa-
vowing any knowledge of the book, I would hope to hear the
magical words “Well, I recommend it highly. Not only did I
learn something I didn’t even know I was interested in, but I
had fun doing it.” Happily, this is still my principal criterion.
So if I chance to drop into the seat beside you on my next flight,
and you enjoyed reading the book as much as I enjoyed writing
it, then perhaps . . .

JLC
Vienna , 1989

ACKNOWLEDGMENTS

There are two characteristics that every inhabitant of that vast
universe of books seems to share. The first is the appearance of
embarrassing typos, literary gaffes, and conceptual errors that
no author’s or editor’s brand of “weedkiller” ever seems able to
eradicate completely. The second is the presence in the book of
the hearts, hands, minds, and souls of others. Like all authors, I
hope that this book will be the exception that proves the rule for
the first universal property, but I’m not placing any bets on it.
As to the second general feature, it pleases me greatly to an-
nounce that this book is no exception. I have been luckier than
most in having had the benefit of the support, encouragement,
opinions, advice, and even services of a large number of people
without whom this project would still be languishing in that
shadowy world of ideas that almost were but aren’t. So it’s both
a pleasure and a privilege for me to bring these unsung heros to
the reading public’s attention here.

Beginning this roll call of honor, the following hardy souls
have through the years acted as sympathetic ears, as well as
intellectual inspirations, in conversations ranging over the
topics of this book and much, much more. In addition, many
of them served as willing guinea pigs for a critical reading
of one or more preliminary versions of the chapters of the book.
So, in no particular order, I thank Karl Sigmund, Clint Per-
kins, Amy Okuma, Manfred Deistler, Gustav Feichtinger,
Lucien Duckstein, Mel Shakun, Jesse Ausubel, Mary McCusker,
David Berlinski, Hugh Miser, Nebojsa Nakicenovic, and Peter
Schwed.

In a book of this sort, keeping the technical details straight is
a job for three men and a boy, not to mention a massive com-
puter database. For their valiant efforts to keep me on the
straight and narrow, technically speaking, I am indebted to Pro-
fessors John Bell, Michael Hart, David Lightfoot, Robert K.
Merton, Michael Ruse, Abdus Salam, John Searle, Robert
Shapiro, John Maynard Smith, and John A. Wheeler. Of
course, these thanks are accompanied by the customary absolu-

XVI

ACKNOWLEDGMENTS

tions for whatever errors of fact and/or interpretation that re-
main.

For specific help far beyond the call of duty, let me also bow
deeply and tip my hat to:

Eddy Loser, librarian extraordinaire, whose genius in tracking
down important, but seemingly inaccessible, references ac-
counts for the unseemly length of the book’s bibliography;

Paul Makin, maestro of the computer terminal, who taught
me what little I know about the ways of computers and their
virtues (and vices) for writing a book;

John Ware, the kind of literary agent every author dreams of:
one who believes in, continually encourages, and works tire-
lessly for his clients;

Bruce Giffords, a copy editor with an eye like a hawk, a mind
like an encyclopedia, and a heart like a lion. If you actually
understand this book, he’s the reason; if not, blame the author;

Alex Grey and Dolores Santoliquido, artists of rare perception
and talent, whose creations grace the dust jackets (AG) and
pages (DS) of the book, illuminating that which was only
darkness before;

Maria Guarnaschelli, the kind of editor every author dreams
of: one who not only protects authors from themselves, but
does it with such grace, humor, talent, artistry, and skill that
the author can still write his book;

Peter de Janosi, my most faithful reader and perceptive
critic, as well as a quintessential role model for that hardy but
vanishing breed, the proverbial educated layman;

Joe Tabacco, Peggy Schmidt, and Teddy Tabacco, friends who
not only provide the most congenial of environments for
an itinerant visitor and the expression of his outlandish opin-
ions on science and life, but who sometimes even agree with
him.

To all these long-suffering friends, my thanks and appreciation
for their many contributions reflected in almost every page of
this work.

ACKNOWLEDGMENTS

XVII

Finally, and most important, heartfelt thanks to my wife,
Vivien, not only for her constant encouragement and support in
all the usual ways too numerous to list, but especially for not
asking to look at the manuscript of this book until it was too late
to do anything about it.

CONTENTS

PREFACE vii

ACKNOWLEDGMENTS xv

1 / FAITH, HOPE, AND ASPERITY 1

BELIEF SYSTEMS, SCIENCE, AND THE

INVENTION OF REALITY 1

WORLD VIEWS IN COLLISION 1

DID YOU SAY SCIENCE! 10

THE NATURAL PHILOSOPHER’S STONES 15

RATIONALITY FOR REALISTS 26

BUDDY, CAN YOU PARADIGM! 38

PHILOSOPHICALLY SPEAKING 46

A TALE OF TWO SUICIDES 48

ON THE FRINGE OR AT THE CUTTING EDGE! 56

THE PULPIT AND THE LAB 62

INTO THE COURTROOM OF BELIEFS 66

2 / A WARM LITTLE POND 68

CLAIM: LIFE AROSE OUT OF NATURAL
PHYSICAL PROCESSES TAKING

PLACE HERE ON EARTH 68

OUT OF THE FIRE AND INTO THE SOUP 68

A CRASH COURSE ON HOW LIFE LIVES 74

POTHOLES ON THE ROAD TO LIFE 84

MONSTERS, HYPERCYCLES, AND NAKED GENIES 88

THE CHICKEN’S STORY 95

LIFE: A TWICE-TOLD TALE 100

ASHES TO ASHES, LIFE FROM DUST 108

IT CAME FROM OUTER SPACE 115

AND GOD CREATED. ..FROM FISH TO GISH 121

THE LOGIC OF LIFE 127

SUMMARY ARGUMENTS 139

BRINGING IN THE VERDICT 140

CONTENTS

3 / IT'S IN THE GENES 143

CLAIM: HUMAN BEHAVIOR PATTERNS
ARE DICTATED PRIMARILY BY THE GENES

NATURE/NURTURE: SENSE OR NONSENSE? 143

NEO-NEO-DARWINISM AND SOCIOBIOLOGY 147

ANIMAL ANTICS 155

THE STRANGE CASE OF ALTRUISM 171

THE GENETIC IMPERATIVE 173

GETTING INTO HER GENES: SEXISM AND SOCIOBIOLOGY 182
CANT VS. KANT 186

SO-SO BIOLOGY 192

CONFLICTING RATIONALITIES AND THE
DILEMMA OF COOPERATION 198

SUMMARY ARGUMENTS 203

BRINGING IN THE VERDICT 205

4 / SPEAKING FOR MYSELF 209

CLAIM: HUMAN LANGUAGE CAPACITY
STEMS FROM A U N I Q U E , I N N A T E
PROPERTY OF THE BRAIN

DUMB DOGS AND CLEVER HANS 209

VERBAL BOTANY AND UNIVERSAL GRAMMAR 213

THE NOAM OF CAMBRIDGE 218

POSITIVELY REINFORCING 232

OUT OF THE MOUTHS OF BABES 237

IT’S ALL A QUESTION OF SEMANTICS 241

SHOOT-OUT AT THE ROYAUMONT CORRAL 249

RULES AND REPRESENTATIONS 253

SUMMARY ARGUMENTS 257

BRINGING IN THE VERDICT 258

5 / THE COGNITIVE ENGINE 261
CLAIM: DIGITAL COMPUTERS CAN,

IN PRINCIPLE, LITERALLY THINK

THE TURING TEST AND THE CHINESE ROOM 261

FORMAL SYSTEMS, MACHINES, AND TRUTHS 268

“STRONG” VS. “WEAK” AI, BRAINS, AND MINDS 285

TOP-DOWN SYMBOL CRUNCHING 290

BOTTOM-UP EMERGENCE 299

PHILOSOPHERS AGAINST: THEY’LL NEVER THINK! 314

THE MORALIST AND THE MYSTIC 324

SUMMARY ARGUMENTS 330

BRINGING IN THE VERDICT 332

CONTENTS xxi

6 / WHERE ARE THEY? 340

CLAIM: THERE EXIST INTELLIGENT BEINGS

IN OUR GALAXY WITH WHOM
WE CAN COMMUNICATE

THE FERMI PARADOX AND PROJECT OZMA 340

THEORETICAL ETI: THE DRAKE EQUATION 343

SLICES OF THE ETI PIE 345

ANTHROPOMORPHISMS, CHAUVINISMS, AND
ETI NUMEROLOGY 362

EXPERIMENTAL SETI: HOW SHOULD WE LISTEN ? 368

WHAT ARE WE LISTENING FORI — THE SYNTAX AND
SEMANTICS OF SETI 373

if > 1: ETI EXISTSI 387

THE SHAPE OF ETIS TO COME 391

ETI? THERE’S NO SUCH THING: AT = 1 397

SUMMARY ARGUMENTS 409

BRINGING IN THE VERDICT 411

7 / HOW REAL IS THE

' ' R E A L W O R L D ' ' ? 414

CLAIM: THERE EXISTS NO OBJECTIVE REALITY
INDEPENDENT OF AN OBSERVER

BUILDING THE STAGE 414

GHOSTS IN THE ATOM 417

MEASUREMENT TO MEANING 429

THE ROMANTIC REALITIES 441

THE DOGWORK REALITIES 456

THE BELL TOLLS FOR LOCALITY 467

IN THE BEGINNING, THE VERY BEGINNING 476

SUMMARY ARGUMENTS 488

BRINGING IN THE VERDICT 489

CONCLUSION /
THEBALANCESHEET 492

ARE HUMANS REALLY SOMETHING SPECIAL?

WHERE DO WE STAND? 492

TO DIG DEEPER 500

INDEX 555

FAITH, HOPE,
AND ASPERITY

BELIEF SYSTEMS, SCIENCE,
AND THE INVENTION OF REALITY

WORLD VIEWS IN COLLISION

On the night of February 24, 1987, Canadian astronomer Ian
Shelton was looking through the telescope at the Las Campanas
Observatory in Chile; what he saw became the scientific event of
the decade in the astronomical world. On that night, Shelton be-
came the first to see the star Sanduleak —69° 202 come to the
end of its cosmic tether in that most spectacular of celestial
fireworks displays, a supernova. According to current astro-
physical wisdom, such events occur when the hydrogen that
fuels the thermonuclear furnaces of stars a little bigger than our

PARADIGMS LOST

sun runs out, allowing the contracting force of gravity to gain
the upper hand over the expanding forces of thermal radiation.
The star’s mass then collapses in on itself until the pressures
build to the point where the star literally blows its top, scatter-
ing most of its mass into the interstellar void, leaving behind a
small, rapidly spinning ball consisting solely of neutrons at an
incredibly high density. In fact, so dense is the material of such
a “neutron star” that one cubic inch of it would weigh more
than a billion tons, and a pinhead’s worth several million. Al-
though many supernovas have been seen in distant galaxies, the
importance of supernova 1987A was twofold: It was the first
time that astronomers had extensive observations of a star
before it became suicidal, and it happened in the Large Magel-
lanic Cloud, a galaxy “only” 170,000 light-years distant — essen-
tially next door on the astronomical scale of things. While
supernovas have been observed from Earth for centuries, going
back at least as far as the Chinese accounts of what is now the
Crab Nebula in a.d. 1054, observation of their neutron star resi-
due dates back only a few years and constitutes one of the major
science stories of the 1960s. Since the discovery of these neutron
stars or, as they as more colloquially termed, pulsars (for “pul-
sating radio sources”) serves as an admirable case study of the
ways of science in the late twentieth century, let’s climb into a
time machine and go back to those exciting days to retrace the
steps leading to this momentous discovery.

The story begins in 1965 with the decision by Jocelyn Bell, a
young woman from Northern Ireland, to seek a doctorate at
Cambridge University in the then-new field of radioastronomy.
As Bell (now Jocelyn Bell Burnell) tells it, she had become fas-
cinated with astronomy as a young girl when her architect fa-
ther was hired to design the observatory in the small Irish town
of Armagh. Unfortunately, even then she saw that a necessary
condition for successful pursuit of the astronomer’s nocturnal
art is to have a night owl’s constitution, easily being able to in-
terchange the normal hours for sleeping and working. Despite
her passion for the stars, in the 1950s her constitutional need for
a good night’s sleep at the normal hours looked like a fatal ob-
stacle to any budding astronomical aspirations. But as luck
would have it, this was the time when Martin Ryle of Cambridge
was developing one of the first telescopes devoted to searching
the skies in the radio rather than visible light part of the electro-

FAITH, HOPE, AND ASPERITY

magnetic spectrum. Since the best time for “seeing” at these fre-
quencies is during the daylight hours, Cambridge was the place
for her, and off she went armed with an undergraduate degree
in physics to work for her Ph.D. in a group led by Anthony
Hewish.

One of the most sacred rules of academic institutions every-
where is that the graduate students perform the slave labor, the
Cambridge Institute of Theoretical Astronomy being a staunch
upholder of this venerable principle. Consequently, Bell spent
her first two years as a graduate student wielding a 20-pound
sledgehammer, helping to construct the radiotelescope that she
would later use to gather the material for her doctoral disserta-
tion. Following completion of the telescope in 1967, team leader
Hewish assigned Bell the thesis topic of measuring the angular
diameter of radio galaxies (quasars) from the way their signals
“twinkled” when seen from Earth due to the solar wind of mate-
rial emitted from the Sun. Her job was to operate the telescope
singlehanded and analyze the output until she accumulated
enough data for a respectable thesis. Since the telescope spewed
out 96 feet of three-track paper each day and covered the entire
sky in four days, Bell’s data analysis activity was hardly less
energy-intensive than building the telescope itself, involving as
it did eyeballing the telescope record and separating the wheat
of true twinkling signals from the chaff of French television,
military radar, aircraft altimeters, and other Earth-based
sources of interference. The telescope was turned on in July
1967 and, not surprisingly, by October she was already 1,000
feet of chart paper behind. It was at this point that the fun,
both galactic and earthly, began.

In one of the 400 feet of chart readings produced with each
scan of the sky, Bell noticed that there was about half an inch of
what she termed “scruff” that resisted classification. She saw
that the scruff was neither twinkling or man-made interference,
and then recalled having seen similar patterns before on another
record from the same part of the sky. Furthermore, she noticed
that the mysterious signals seemed to be appearing periodically
on sidereal time of twenty-three hours, fifty-six minutes, i.e., the
time needed for a given location on Earth to return to the same
position relative to the fixed stars (the sidereal day is four min-
utes shorter than the terrestrial day due to the Earth’s orbital
motion about the Sun).

PARADIGMS LOST

At this juncture Bell discussed the signals with Hewish, and
they decided to look at them again on a faster recorder that
would allow them to pick out more detail. This recorder was oc-
cupied at the moment, so they had to wait until mid-November
to make the new reading. As so often happens in life, just when
you want a taxi (or a cop) there’s not one to be found anywhere;
astronomical anomalies are similar, and Bell had to wait several
weeks before she could reacquire the odd signal. Imagine her
surprise when she finally found it again and discovered that it
was pulsating at the metronomic rate of almost exactly 1 V& sec-
onds. She immediately phoned Hewish, who promptly dismissed
the signals as man-made in light of their extreme regularity.
However, an Earth-based source would keep terrestrial time, not
sidereal, casting a very dark shadow over Hewish’s offhand con-
clusion. But the fastest variable star then known had a period of
one third of a day, and it was difficult to conceive of what kind
of star would rotate in little more than a second.

The first attempt to reconcile these conflicting facts was to
conjecture that the observations were radar signals bouncing off
the Moon, or a satellite in an odd orbit. But such an explanation
didn’t wash, and since only astronomers and the stars keep side-
real time, Hewish thought that perhaps some other observatory
had a program under way that would account for the unusual
signals. His queries to other radioastronomers turned up no
such program. The next trial explanation was the LGM Hypoth-
esis, postulating that the signals were intelligent communica-
tions from “little green men.” As a test of this conjecture,
Hewish calculated the Doppler shift of the pulses assuming that
the LGM would be on a planet, and that the planet’s orbital
movement around its star would create a clustering of the pulses
as the planet moved toward Earth and a spacing-out of the sig-
nals as it moved away. This explanation also came a cropper
when the only Doppler shift noted was that due to the Earth’s
motion around our sun. At this point, theory gave way to an-
other observation, which definitively settled the matter.

Just before leaving for her Christmas holiday in December
1967, Bell was working late one night analyzing a record from a
different part of the sky. She noticed some more scruff that
looked remarkably similar to that of the LGM signal. As seren-
dipity would have it, the telescope was due to scan that part of
the sky again that very night, and she luckily got a strong read-

FAITH, HOPE, AND ASPERITY

ing showing an extremely regular train of pulses coming in at
the rate of about lVi seconds per pulse. Since another rule of
graduate student life is that you don’t telephone your professor
at 3 a.m. (at least you don’t if you value finishing your degree
program), Bell just dropped the recording on Hewish’s desk
with a note asking him to keep the recorder going over the vaca-
tion period, and left for her holiday. Hewish himself then made
a recording in mid- January confirming the second source,
thereby removing the LGM hypothesis from further considera-
tion on the grounds that it was extremely unlikely that there
could be two groups of LGM trying to signal us on different
frequencies at the same time. So when Bell returned from her
Christmas break, she had two important problems to deal with:
(1) there was more than one pulsar, and (2) it was time to start
writing up a thesis describing her original work on the angular
diameter of quasars (although it ultimately contained an appen-
dix describing the pulsar observations, too).

Forced into accepting that the sources of these pulses were
some sort of stellar phenomena, Hewish, Bell, and three others
from the Cambridge team coauthored the first paper on the sub-
ject, which was published in February 1968, and which vacil-
lated between identifying the sources as neutron stars and as
white dwarfs, the kind of object our own sun will contract into a
few billion years from now. Six months later, the astrophysical
community accepted Thomas Gold’s interpretation that they
were neutron stars as being the only plausible explanation fit-
ting all the observations. This proposal followed up a theoretical
suggestion that Fritz Zwicky and Walter Baade made in 1934.
The general picture of how a neutron star acts to produce the
observations seen by Bell and Hewish is shown in Figure 1.1.
While the scientific excitement ended here, the story was still far
from over.

In 1974 the Nobel Committee awarded its prize in physics for
the first time to astronomers, citing Martin Ryle and Anthony
Hewish for their “decisive work in the discovery of pulsars.”
Not a word was said about the actual discoverer of pulsars,
Jocelyn Bell! Shortly after the award ceremony in December,
another member of the Cambridge astronomical group, Fred
Hoyle, said in a speech in Montreal that Bell’s findings had been
kept secret for six months while her supervisors “were busily
pinching the discovery from the girl, or that was what it

PARADIGMS LOST

Beam of

FIGURE 1 . 1 A pulsar in action

amounted to.” Hewish admitted that he was “angry” over
Hoyle’s allegation, calling it “untrue,” and noting that “Jocelyn
was a jolly good girl but she was just doing her job. ... If she
hadn’t noticed it, it would have been negligent.” He went on to
state that she had made the discovery using his telescope, under
his instructions, making a sky survey that he had initiated.
Other astronomers were less certain. The historical fact re-
mained that Bell was the first person who had recognized the
pulsar signals, and in fact she and Hewish, presumed to have
shared equally in the work by the exacting standards of the
Franklin Institute’s awards committee, were jointly awarded
the institute’s prestigious Michelson Medal in 1973 for the dis-
covery.

Personally, I’ve always felt that Hollywood missed a good bet

FAITH, HOPE, AND ASPERITY

by not putting this story on film, showing an upset, slightly
bookish Jane Fonda or Meryl Streep look-alike publicly de-
nouncing a suave, but faintly sinister, James Mason-ish profes-
sor on the steps of the Stockholm City Hall for casting her and
her contribution aside in pursuit of personal fame and glory.
Unfortunately for Hollywood, real life as usual had quite a dif-
ferent ending in mind. In response to the various claims and
counterclaims, Jocelyn Bell had the last word when she stated
that Hoyle “has overstated the case so as to be incorrect.” But
still, given the proclivity of the film industry for warping and
distorting reality in pursuit of art and entertainment, not to
mention hard cash, maybe there’s hope yet for realization of my
vision. In any case, the entire pulsar episode serves as a sterling
example of the bright side of the folkways, mores, and byways
of contemporary scientific life. For a look at the dark side, let’s
return to our time machine and go back a few more years to
examine another tempest in the astrophysical teapot.

In the writings of Plato and Herodotus we find the assertion
that the Sun now rises where it once set. How could they make
such a bizarre claim? And why do so many cultures have legends
of global floods, manna from heaven, darkness on the Earth, and
other such strange phenomena? In 1950 the Macmillan Publish-
ing Company put out the volume Worlds in Collision by a Rus-
sian-born psychoanalyst, Immanuel Velikovsky, who purported
to explain these and many other phenomena as the result of a
series of celestial cataclysms taking place during historical
times. This book so enraged the scientific community that Mac-
millan, under pressure of a boycott of its textbook division,
handed the best-selling project over to Doubleday and fired the
editor responsible for dealing with the manuscript. It’s instruc-
tive to examine Velikovsky ’s claims and methods as an example
of the sort of thing that sends the scientific establishment into
apoplectic fits.

The gist of Velikovsky ’s argument is that a large comet was
expelled from Jupiter sometime around the year 1500 b.c. This
comet passed very close to us, with its tail touching the Earth
and causing a rain of petroleum, as well as darkening the sky
for several days with its dust and debris. In addition, the
Earth’s rotation rate was slowed down by the comet, resulting in
earthquakes, hurricanes, tidal waves, and a variety of other dra-

PARADIGMS LOST

matic environmental shenanigans. Electrical discharges between
the Earth and the comet caused a reversal of the Earth’s mag-
netic field, the polar regions shifted, and the Earth’s axis of ro-
tation was altered, resulting in a change in the order of the
seasons. Furthermore, the Earth was pushed into a larger orbit,
lengthening the year to 360 days.

Velikovsky correlates this first pass of the comet with the Ex-
odus of the Israelites from Egypt, claiming that the plagues of
blood, vermin, and hail noted in the Bible were the result of the
Earth s contact with the comet’s tail. He also explains the part-
ing of the waters of the Red Sea as being due to the stopping of
the Earth’s rotation, and that the manna from heaven sustain-
ing the Israelites in the desert was composed of carbohydrates
from the comet. Worlds in Collision then asserts a second passage
of the comet fifty-two years later, this time interfering with the
Earth s rotation just at the time when Joshua commanded the
Sun to stand still. And what does Velikovsky say about the iden-
tity of this celestial molester? He claims that the comet is now
what we call the planet Venus! But the story doesn’t end there.

In Velikovsky ’s scenario there was another close cometary en-
counter around the year 800 b.c., this time with the planet Mars.
This near collision knocked Mars out of its orbit, bringing it
close to the Earth on at least three occasions. These near misses
shifted the Earth’s orbit even further away from the Sun,
bringing about the current year of 365 V* days. At this point, all
three planets settled into their current positions, thus folding up
the tent on Velikovsky’s celestial circus.

One might well inquire as to what kinds of arguments and
methods Velikovsky employed to explain these catastrophic go-
ings-on. Fundamentally, Worlds in Collision is based upon an-
cient manuscripts, legends, and traditions. In a later volume,
Earth in Upheaval, he cites evidence such as the existence of coal
beds in Antarctica, rock formations with reversed magnetic po-
larity, fossil beds containing animals from both desert and for-
est, as well as other geological and paleontological facts. The
cometary origin of Venus also gave rise to Velikovsky’s specula-
tions that Venus was hot and that the material for the comet
had originally been expelled from Jupiter, leaving behind what
we now know as the giant Red Spot.

It probably goes without saying that mainline astronomers,
geologists, astrophysicists, and paleontologists speak with one

FAITH, HOPE, AND ASPERITY

loud voice in their condemnation of both Velikovsky’s methods
and his conclusions. While his work represents an imposing
piece of sustained scholarship, there are just too many inconsis-
tencies in far too much of his historical, archaeological, astro-
nomical, and physical data to take the arguments seriously. For
instance, while it did turn out that Venus was scorchingly hot,
just as Velikovsky had predicted, this is almost certainly due to
an atmospheric “greenhouse effect” and not to any kind of come-
tary origin. Furthermore, the atmosphere of Venus is almost to-
tally devoid of the hydrocarbons that Velikovsky claimed would
be found as its main constituents. Moreover, the surface of
Venus appears to be over 1 billion years old, instead of just a
few thousand years as predicted by Velikovsky. For these rea-
sons and many more, Velikovsky’s vision of the solar system has
now been relegated to that corner of the scientific attic where sit
ancient astronauts, the Piltdown man, phrenologists, astrolo-
gers, and all the other playmates of the pseudoscientist.

Despite the truly devastating holes in his theory, Velikovsky
died in November 1979 convinced that he had been the victor in
his war against the Brahmins of science. And, in fact, his ideas
live on to this day in some circles. In our quest here to uncover
the essence of what constitutes “scientific” knowledge, it’s worth
taking a moment to examine the pulsar and Worlds in Collision
theories as antipodes of the spectrum of what is commonly
termed scientific research.

At first glance, there appear to be a number of similarities
between the work of Bell and Hewish on pulsars and that of
Velikovsky: unexplained astronomical phenomena, conjectures
and refutations of various theoretical explanations, a physically
unobservable explanation interpreted to fit the observations —
even a public controversy over some sociological aspects of the
way the world of science goes about distributing its accolades.
With these points of contact, why is it that the scientific commu-
nity chose to reward Hewish with its highest honor, the Nobel
Prize, while at the same time vilifying Velikovsky and dismiss-
ing him as what could charitably be termed a misguided crank?
J ust what was it exactly about the pulsar work that made it the
height of respectability and was so obviously lacking in the ef-
forts of Velikovsky?

The long and proper answer to the question will occupy us for

PARADIGMS LOST

much of the remainder of this chapter; the short answer is that,
by common consensus in the scientific community, certain stan-
dards have been set for what constitutes acceptable evidence and
methods, with the pulsar work adhering to them while Veli-
kovsky’s did not. The central point for us in this volume is the
degree to which those commonly accepted standards generate
real rather than virtual knowledge of the universe in itself. Put
another way, do the methods and standards of science produce a
brand of knowledge that is somehow more certain or of higher
intrinsic pedigree than the methods and standards of other seek-
ers after truth like Velikovsky? The first step toward a resolu-
tion of this overarching question is to address a different
question: Just what does constitute the practice of “science” as
that term is commonly used in today’s world?

DID YOU SAY SCIENCE?

Back in the days when I still attended cocktail parties, the most
awkward situations always arose at those odd moments when the
music stopped and social convention dictated that I make some
feeble effort to “mix.” Generally at these times, life conspired to
place me next to some slightly frenetic, upwardly mobile yuppie
type suffering from an overdose of adolescent enthusiasm for
drinking deeply from the brackish waters of life, not to mention
our host’s bar. Inevitably such encounters began with the ques-
tion “What do you do?” Resisting the temptation to reply, “Ah,
yes, the eternal question,” or give some other equally sophomoric
response, in the early going I used to answer honestly that “I’m
a mathematician.” The reactions to this bit of ill-advised candor
fell into one of two categories: a petulant pout followed by the
curious compliment that “I was always terrible in math,” or
what was even worse, a bright smile and the remark “Oh, you’d
love my uncle. He’s an accountant.” Being a slow learner, I
needed some time to realize that such frank confessions of pro-
fessional perversion were not the road to success on the cocktail-
and-corn-chip circuit. So I began experimenting with other, less
esoteric replies: “I’m an electrical engineer, a chemist, an
agronomist [“What’s that?”], a scientist.” The results could
hardly have been worse if I’d claimed to have been a psychia-
trist, an undertaker, or, heaven forbid, some back-slapping
politico type. Finally, I hit upon the winning solution of just

FAITH, HOPE, AND ASPERITY

saying that I was an unemployed tennis coach, at which point
my Social Interaction Index shot up like a Minuteman missile.
But the sad conclusion to be reached from this very statistically
insignificant sample is that there is a wide variety of gross mis-
conceptions and nontrivial misunderstandings floating around,
even among the educated public, as to the nature of both scien-
tists and the ways in which they spend their days (and nights!).

Trying to distill the essence out of the aforementioned encoun-
ters, I eventually came to the surprising realization that the
term science seems to be used interchangeably in general conver-
sation in at least three quite distinct and inequivalent ways:

Science -

a set of facts and a set of theories that explain the
facts

a particular approach, the scientific method

whatever’s being done by institutions carrying on
“scientific” activity

As a general rule, the nonscientific public usually opts for the
third interpretation, occasionally the first, but virtually never
the second — just the opposite ordering from that given by the
scientific community itself. It’s no wonder C. P. Snow could de-
velop a lengthy essay on the “two cultures.”

The fundamental misunderstanding on the public’s part of
what constitutes a “scientific” activity gives rise to an array of
subsidiary misperceptions about the goals of science and the way
scientists go about their business of trying to achieve them. Let
me list just a few of the more important popular fictions:

• The primary goal of science is the accumulation of facts. Unfortu-
nately, the mere cataloguing of data is not enough; we also
require some overall organizing principles and a relationship
between these principles and the data. Actually, for scientists
the more reliable a fact is, the more trivial and unimportant it
becomes. For instance, the atomic weight of carbon can confi-
dently be given as 12.011 atomic units. Yet this fact is basi-
cally just a curiosity until it’s correlated with similar facts
about the other chemical elements, using the laws and theories
of chemistry and physics.

• Science distorts reality and can’t do justice to the fullness of human
experience. Every human undertaking must somehow pick and
choose as to what aspects of reality to omit in order to probe
other aspects of the world. In this regard science is no differ-

PARADIGMS LOST

ent from religion, art, literature, mysticism, or any of its
other competitors in the reality-generation business.

• Scientific knowledge is truth. Science is not in the business of
providing ultimate explanations. Every scientific law or the-
ory is subject to modification; there are no universal, absolute,
unchangeable “truths” in science.

• Science is concerned primarily with solving practical and social prob-
lems. I can’t think of a single statement about science that could
be further from the actual case. For most scientists, science is
a game played for understanding, not for obtaining practical
information about how to build a better radio, mix more nutri-
tious dog food or iron out the wrinkles of middle-aged dowag-
ers. In fact, this “science = technology” misperception is so
pervasive that it merits a few additional words all its own.

Some time back, I had the enervating experience of working
for a man who suffered from the delusion that doing science
meant finding answers to practical problems posed by industrial-
ists, government policymakers, and other dreamers, schemers,
and so-called men of affairs. One conversation that I ruefully
recall involved my temerarious claim that if you focus attention
on finding well-defined answers, then you’re not doing research,
at least not scientific research. Research involves ideas, not an-
swers. In my view, what counted was developing a deep under-
standing of the question itself; whatever “answers” there might
be would then follow as corollaries of this insight into the real
nature of the question. A solution itself is not the ultimate goal;
what’s important is understanding why an answer is possible at
all, and why it takes the form that it does. The point I was mak-
ing was that technological advancement and the acquisition of
scientific knowledge have only the feeblest points of contact with
each other. Technology is primarily engineering, and new tech-
nologies come more from fighting with physical reality than
from scientific theories. Besides, it’s not clear that new technolo-
gies give us a better understanding of nature anyway, e.g., mod-
ern medicine vis-a-vis Chinese acupuncture.

The moral of the foregoing little tale is that even many people
who practice under the rubric of what in the vernacular is called
a scientist hold to a view of science and scientific work that at
best falls into the third category noted earlier, which we might
compactly describe as “the General Electric Syndrome.” That is,
if GE is doing it, it must be science. Well if GE is doing it, it

FAITH, HOPE, AND ASPERITY

probably isn’t science, at least not the kind of science that most
members of the global scientific community would recognize. It
may be high-grade technology or world-class engineering or even
pathbreaking developmental research, but definitely not science.
I hasten to point out here that this observation is in no way
intended to minimize the truly outstanding and genuine scien-
tific work that is carried out at places like GE, IBM, Bell Labs,
Exxon, and so on. But it’s not the real science going on in these
corporate research labs that members of the public have in mind
when they think of, say, IBM. What comes to mind is comput-
ers, typewriters, and all the other office paraphernalia that car-
ries the IBM logo and that people use in their day-to-day affairs.
The development of these gadgets is the main business of such
an institution, and that development is definitely not science; it’s
technology. Now let’s get back on course and examine just what
it is that does constitute science as it’s seen by the scientists
themselves.

Paradoxically, scientists usually think of science as one area
of life in which ideologies play no role. Nevertheless, there is a
collection of beliefs and ideals about the practice of science that
the scientific community clings to with such universal tenacity
that it’s difficult to describe it as anything other than an ideol-
ogy— the ideology of science. The scientific ideology is a mixture
of logical, historical, and sociological ideals about how science
should operate in a Panglossian world, and rests upon the fol-
lowing pillars:

• The logical structure of science: This pillar represents what
many of us learn in our early schooling about the procedures
followed in science. Here we find the sequence:

Observations/Facts

Hypothesis

Experiment

Laws

Theory

To many, this diagram represents the essence of what we think
of as the scientific method. Observations give rise to conjectures
and hypotheses, which in turn are checked out by performing

PARADIGMS LOST

experiments. If the experiments don’t confirm the hypothesis,
then new hypotheses are formed, just as in the pulsar work de-
scribed earlier. Those hypotheses that survive are encapsulated
into empirical relationships, or laws, which in turn are embed-
ded in larger explanatory theories. It is this sequence of steps
that’s been the focus of most of the philosophical analyses of the
process of science, as we shall discuss in detail later. However, to
the practicing scientist there is much more to the scientific enter-
prise than mere philosophy.

• Verifiability of claims: Science is a public undertaking with
many filters that a claim must pass through before it’s ac-
cepted as part of the current conventional wisdom. Two of the
most important are the refereeing process for scientific arti-
cles and the repeatability of experimental results. Before a
reputable scientific journal will publish a research announce-
ment, it’s sent out for review to other workers in the field, not
only as insurance that the results are correct, but also to sub-
stantiate their significance within the framework of current
knowledge in the area. In a similar manner, published work is
supposed to report all the details of the investigator’s experi-
mental setup so that any interested party can, in principle,
repeat the experiment and try to replicate the reported results.
Thus, in the utopian world where the scientific ideology reigns,
refereeing and repeatability keep the scientific process (and
the scientist) honest.

• Peer review: The modern scientist is in much the same situa-
tion as the artisan of the Renaissance, at least when it comes
to needing a patron to finance pursuit of the muse. The only
difference is that nowadays everyone has the same patron — the
federal government. As a result, most funds are allocated by
federal agencies, making liberal use of the so-called peer re-
view process. This involves committees of experts from the
various fields getting together and recommending to the fund-
ing agencies those projects and those scholars whose work they
feel merits support. According to the ideology, this process en-
sures that money is channeled to those ideas, institutions, and
individuals showing the clearest evidence of being able to do
something productive with it.

Given the highly egalitarian, logical, meritocratic nature of
the scientific ideology, it comes as no surprise that many scien-
tists accept it as at least a very close approximation to the way

FAITH, HOPE, AND ASPERITY

science really is. I’ll defer detailed consideration of this point to
a later section. At the moment let me just remark that a neutral
skeptic would almost certainly raise an eyebrow or two over the
rather obvious fact that the conventional ideology focuses en-
tirely upon the process of science, leaving aside all considerations
of the motives and needs of the scientists themselves. The degree
to which this omission casts a cloud over the rosy picture
painted above will occupy our attention throughout the book.
For now, let’s stick to the scheme above and turn the spotlight
on the cognitive structure of science, in an attempt to get back
to the questions of just what kind of knowledge the process of
science is able to offer us about the nature of the world as it is,
and whether that kind of knowledge is in some way superior to
any other kind.

THE NATURAL PHILOSOPHER’S STONES

The issue before the house for the next couple of sections is con-
sideration of the dual questions:

Do scientific theories in any sense tell us about the way the
world is?

Does science have anything like a method for creating and/or
evaluating theories?

Since all theories must necessarily be expressed in some kind of
language (natural, symbolic, mathematical), the first question
takes us into the province of the philosophy of language as a tool
for representing reality. The second question deals more with
science per se, forcing us to confront the natural query “What’s
so special about science?” In other words, why should we believe
that scientific knowledge is any more correct or reliable than any
other sort? So our short-term objectives are to explore the ques-
tion marks in the following diagram:

Scientific theory X> Objective reality

T ?

Scientific methods

To address these two foundational question marks, it will be nec-
essary for us to dip briefly into the work of several twentieth-

PARADIGMS LOST

century philosophers of language and science. But before delv-
ing into the ideas of these thinkers, let’s first go back a couple of
millennia and fix our attention on some of the pivotal ideas of
the ancient Greeks that ultimately led to the confused state we
find ourselves in today.

In his last will and testament, Aristotle offers the following
logical sequence of steps — i.e., an algorithm — for disposition of
his estate. Until his chosen son-in-law, Nicanor, came of age, the
estate was to be managed by three executors. If Nicanor died
prior to the time when Aristotle’s daughter, Pythias, would be
old enough to marry him, then Theophrastus was to step in and
fill Nicanor’s designated role. But if Pythias married someone
else who, in the opinion of the executors, didn’t disgrace Aris-
totle’s name, then she was given permission to use the family
ancestral home at Stagira, which was then to be furnished to her
satisfaction by the executors. Even after death, Aristotle leaves
no stone unturned and no possibility unaccounted for — just the
kind of detailed, step-by-step prescription that we might have
expected from the man who invented the idea of formal logical
deduction.

For Aristotle, the procedure for uncovering the truth of
things was to postulate premises, then use the now-familiar
rules of logical deduction to derive the consequences implicit in
the premises. The classical example of this procedure, which
we’re all familiar with from Philosophy 101, is:

Premise I: All men are mortal.

Premise II: Socrates is a man.

„ . 4

Conclusion: Socrates is mortal.

Note that nothing is said here about the actual truth or falsity
of the premises. Maybe some men are not mortal or maybe Soc-
rates is really a woman or a hermaphrodite or whatever. Physi-
cal reality and truth play no role in the deductive method; the
premises are assumed to be true, with the conclusion following
from this assumption.

Prior to Aristotle the traditional means for structuring expe-
rience was the myth, a term deriving from the Greek mythos,
meaning “word,” in the sense that it is the definitive statement
on the subject. A myth presents itself as an authoritative ac-
count of the facts that is not to be questioned, however strange

FAITH, HOPE, AND ASPERITY

it may seem. According to the famous mythologist Joseph Camp-
bell, myths serve several functions:

• Metaphysical: Myths awaken and maintain an “experience of
awe, humility and respect” in recognition of the ultimate mys-
teries of life and the universe.

• Cosmological: Myths provide an image of the universe and ex-
planations for how it works.

• Social: Myths validate and help maintain an established social
order.

• Psychological: Myths support the “centering and harmoniza-
tion of the individual.”

Myths need be neither true nor false, just useful fictions; how-
ever, they are not the kind of fiction that has entertainment
value alone, and makes no pretensions to truth. Religion, as we
shall see later, goes one step further than the useful fiction of a
myth by making assertions about what is indeed the case. It is at
this point that the age-old conflict between science and religion
starts to take off.

To illustrate the use of myths, imagine a band of prehistoric
hunters who have spent several days stalking a herd of mam-
moths. Just at the moment of truth when they’ve laid their am-
bush and are about to attack, a thunderbolt from the sky comes
flashing down, scattering the herd and undoing all the hunters’
carefully laid plans. Somehow it’s comforting at such times for
the hunters to have a belief system that provides some explana-
tion for what would otherwise seem a capricious whim of the
cosmos. A myth provides such a system of beliefs by offering a
scheme by which to order and explain the thunderbolt. Perhaps
the gods were angry because they had not been properly hon-
ored, or maybe the spirits of dead mammoths from the past
warned their living brethren, or it might have been that the
hunters hadn’t approached from the right direction. Whatever,
the important point is that the myth serves as a schemata
whereby the events of daily life can be given an interpretation in
terms of mysterious forces and beings whose powers transcend
lowly human concerns. Aristotle began the process of replacing
myth with what has now come to be termed science.

The opposite side of the reality coin from mythos is logos, the
Greek term for an account whose truth can be demonstrated and
debated. It is this kind of truth that Aristotle was trying to

PARADIGMS LOST

grasp when he developed logos into “logic” by use of the process
of deduction. One of the main uses of myths as outlined above is
to provide an explanation of how real-world events work. In ev-
eryday speech, an “explanation” is usually taken to be the an-
swer to a question that begins “Why?” Such answers inevitably
begin with “Because,” and the question and answer together
constitute what we generally call a statement of cause and effect.
Thus, “Why is the sky blue?” is answered with “Because the air
molecules absorb all frequencies of visible light except those in
the blue part of the spectrum.” And “Why does water boil at
100°C (at sea level)?” is answered by “Because at that tempera-
ture the thermal motion of the water molecules is able to over-
come the external atmospheric pressure” — cause and effect,
stimulus-response. The method of logical deduction is Aristotle’s
theoretical, or some might say mathematical, counterpart to the
explanation of physical happenings by cause and effect.

In his Physics, Aristotle attempted to combine the purely logi-
cal method of deduction with his ideas about the nature of phys-
ical reality in order to draw conclusions about the way the world
really works. In Aristotle’s view physical matter was composed
of three things: qualities, form, and spirit. He felt that there
was only one kind of matter, which could take many forms, the
fundamental forms being air, earth, fire, and water. Because
these four fundamental forms were not elements in any sense in
which we might understand that term, they could be trans-
formed into each other. To illustrate, this scheme gave rise to
what today we might term Aristotle’s version of the hydrologic
cycle: The Sun’s heat changes water into air; heat rises, so the
heat in this air pulls the rest of it up to the skies; the heat then
leaves the vapor, which becomes progressively more watery
again, and this process results in cloud formation. There ensues
a positive feedback effect in which the more watery the cloud,
the more the water drives away its opposite, the heat. Thus, the
cloud gets colder and contracts. The contraction then restores
true wateriness to the water, which falls as rain or, if the cloud’s
heat has now fallen below the freezing point, hail or snow. So we
see here the relentless chain of cause and effect being employed
to “explain” the observed behavior of water, air, heat, rain, and
snow. What’s amazing about the whole setup is how all the
wrong reasons somehow combine to produce something remark-
ably close to the way things really do work!

FAITH, HOPE, AND ASPERITY

For almost two thousand years Aristotelian logic and physics
served as the “science” of the time, explaining various aspects
of nature, body, and mind by logical consequences of assump-
tions of the foregoing type about the nature of matter. Oddly
enough, despite Aristotle’s main occupation as an observational
biologist, the biggest flaw in his entire world picture was that he
advocated no experiments or even use of observations to serve as
a check on the validity of his underlying premises. Basically, his
was an epistemology in which one inferred specific instances
(conclusions) from general observations (premises). It was not
until the work of Francis Bacon in the seventeenth century that
someone had the courage to challenge the authority of Aristotle
and suggest turning the situation around, i.e., trying to infer
general instances from specific observations.

Bacon’s argument was that if one wants to come to grips with
the way the world really is, it’s necessary to begin the investiga-
tion with the facts of life rather than prejudices about what
those facts might be. Thus followed the principle of induction,
whereby conclusions about future events are drawn on the basis
of repeated past observations. Such an approach is just what we
might come to expect from a man who was not only a philoso-
pher, but also a lawyer who rose to the post of lord chancellor of
England before being dismissed for taking a bribe (an indica-
tion, perhaps, that the current dubious ethical state of the legal,
financial, and political professions are not late-twentieth-century
aberrations, after all). In Bacon’s view of things, if we observe
the Sun rising in the east for fifty consecutive days, then we can
predict that it will rise in the east on day 51. And the longer we
observe such regular behavior, the more confidently we can
speak about its continuation. In a nutshell, this is the method of
induction — lots of individual observations eventually resulting
in the inductive leap to a general conclusion.

On the one hand, it’s satisfying to have a method that takes
into account what Nature is actually doing; on the other hand,
why should such a procedure provide reliable information about
the way things work? On what grounds can I be certain that
every time I put water into my ice-cube trays and leave them in
the freezer for a few hours I’ll soon have ice for my scotch on
the rocks? Just because it’s always happened this way before,
does that give me any assurance that today’s drink will have the
customary satisfying “clink”? The short answer is that there’s

PARADIGMS LOST

absolutely; no justification at all for my concluding that I’ll soon
be enjoying a scotch on the rocks and not a scotch and water.
This is the Problem of Induction: Why should induction work?
Why is it a reliable guide to the future?

To illustrate the Problem of Induction, consider the following
exchange:

woman: Professor, professor. You must help me. My husband
uses an inductive argument to justify the use of inductive ar-
guments.

professor humE: That’s terrible. How long has he acted this
way?

woman: As long as I can remember.

humE: Then why didn’t you see me sooner?

woman: I would have, but we needed (the conclusions of) the

inductive arguments.

humE: I’m afraid I need them too.

Philosophers beginning with Hume have grappled with this
problem, and I’ll consider some of their conclusions in the next
section. For now we leave it as a gaping hole in the attempt to
repair the difficulties in Aristotle by introducing actual observa-
tions into the creation of a world view.

Galileo and Newton are the last two supporting actors in our
cursory sketch of developments leading up to the modern era of
scientific “truth.” Galileo was a contemporary of Francis
Bacon, and although there appears to be no record of direct con-
tact between the two, there is a clear connection between the idea
of Nature as the arbiter of what’s what as advocated by Bacon,
and Galileo’s refinement of the idea by instituting the notion of
a controlled experiment. In effect, Galileo said that if you have a
theory about how some phenomenon works, you must construct
an experiment in which all the variables except the one you’re
interested in are controllable. Then, by fixing the controlled vari-
ables, you can measure the variable of interest, thereby checking
your theoretical hypothesis against the supreme court of obser-
vation. Thus follows the oft-recounted legend (for which there’s
not a shred of documentary evidence) of his experiment of drop-
ping two different weights from the Leaning Tower of Pisa, and
measuring their respective rates of fall as a “laboratory test” of
the hypothesis that objects fall at a uniform rate in the absence
of air resistance, irrespective of their mass.

FAITH, HOPE, AND ASPERITY

Newton added the idea of the description of nature in mathe-
matical terms — the keystone in the arch of scientific knowledge
whose foundations were laid by Aristotle. More than his remark-
able experimental results in optics, mechanics, and chemistry,
Newton’s legacy as writ large in his Principia is the idea of what
we would today call the mathematical model. Newton showed not
only how to “encode” Bacon and Galileo’s world of observation
into mathematical form, but also invented the method (calculus)
for using the mathematical machinery to grind out theorems
that could be “decoded” into new implied statements about Na-
ture. The essence of this procedure is depicted in Figure 1.2,
where the physical system to be modeled (e.g., the solar system,
an electrical circuit, or whatever) is on the left, while the formal
mathematical system that represents it appears on the right.
Also on the left is our earlier notion of causality, represented as
a property of the physical system in which certain parts of the
system exert influences “causing” things to happen elsewhere in
the system. The term implication is used on the right to represent
either the process of Aristotelian deduction or that of Baconian
induction as the means of proving mathematical statements to be
logically correct. These statements are usually called theorems
and follow from axioms and the above logical rules of inference.
The set of implications is the logical counterpart of the physical
causality noted on the left side of the diagram. These implied
statements are then interpreted — i.e., decoded — into assertions
about the way the material system really is.

With the ideas of deduction, induction, observation, and ex-
periment welded together by the symbolic formalism of mathe-
matics, the stage is now set for a brief account of the alphabet
by which modern science tries to inscribe the secrets of nature.
The main letters in this alphabet are facts/observations, laws,
theories, and models. Let’s take a look at what each of these con-
cepts means in the context of modern science.

In Dickens’s tale Hard Times, the schoolmaster Thomas Grad-
grind opens the story with the statement “Now, what I want is,
Facts. Teach these boys and girls nothing but Facts. Facts alone
are wanted in life. Plant nothing else, and root out everything
else. You can only form the minds of reasoning animals upon
Facts: nothing else will ever be of any service to them. . . . Stick
to Facts, Sir!” While Gradgrind is hardly a role model of the

PARADIGMS LOST

decoding

FIGURE 1.2 Newton’s scheme for mathematical modeling

kindly, scholarly schoolmaster, his view forms the starting point
of what many think of as constituting “reality”: the world we
can see, touch, smell, and hear; the world of Facts. But this com-
monsense view is only the starting point for a scientific investi-
gation of Nature’s scheme of things. As noted earlier, isolated
facts are useless curiosities until they are put together with
other facts into some kind of pattern. This requires the develop-
ment of laws.

Suppose we do the following experiment : Take a long cylinder
with a movable piston and fill it with gas (e.g., one of the cylinders
in the motor of your car). Imagine now that we move the piston to
various positions, and for each position measure the pressure
that the enclosed gas exerts upon the walls of the cylinder. Fur-
ther, suppose that after performing many such measurements,
we note that whenever the volume of the cylinder is decreased by
a certain fraction, the pressure increases by the same fraction;
similarly, if we increase the volume by a fraction A by letting
the piston rise, we find that the pressure decreases by the same
amount A . By an inductive argument, after many repetitions of
this experiment we would eventually conjecture (hypothesize)
that there is a direct relationship between the pressure and the
volume of the gas in the cylinder. Specifically, we would proba-
bly assert that the pressure P is inversely proportional to the
volume V. And if we were mathematically inclined, we would
compactly write this relationship as PV = k, where k is a con-
stant determined by the nature of the particular gas and the
units of measurement being used. This relationship is an exam-
ple of what is called an empirical law. The law enables us to sum-
marize a large number of individual facts (the results of the
individual experiments) in one general statement.

FAITH, HOPE, AND ASPERITY

The characteristic properties of laws of the foregoing type are
that they:

1. are about kinds of events (experiments involving the pres-
sures and volumes of gases in cylinders), not about any sin-
gular event (a particular experiment with a particular
cylinder using a particular gas);

2. show a functional relationship between two or more kinds of
events;

3. are supported by a large amount of experimental data contain-
ing little or no disconfirming evidence;

4. are applicable to different events (other types of gases and/or
cylinders).

It’s important to observe here that there are many different
types of laws, not all of which are scientific. The reader might
like to try to distinguish among the following in regard to their
scientific character: parking regulations, the Ten Command-
ments, the Law of Conservation of Energy, the Law of the Ex-
cluded Middle.

Useful as it is, the above pressure-volume relationship
(Boyle’s Law) still doesn’t tell us why an increase in pressure is
linked with a decrease in volume. For this we need a theory of
gases. An explanation for Boyle’s Law can be obtained only if
we invoke the atomic nature of the gas, and think of it as being
composed of a large number of little “billiard balls” randomly
moving about, occasionally colliding with each other and with
the walls of the cylinder. Newtonian mechanics describes the mo-
tion of each such ball, and by combining their individual mo-
tions we can in principle calculate the pressure on the container
walls by determining how many balls are colliding with the walls
at each instant, and the strength of each such collision. With
this picture in mind, it’s easy to see why when the volume of the
cylinder is halved, the pressure doubles. Since the cylinder’s sur-
face area has been cut in half, the likelihood that a randomly
moving ball will collide with the wall doubles. Newton’s laws of
mechanical motion in the context of this gas situation form the
basis for what is termed the Kinetic Theory of Gases, a frame-
work that enables us to explain Boyle’s Law.

The characteristic feature of a theory is that it offers a means
of relating the laws describing a class of events to a framework
and a set of principles described in terms differing from those

PARADIGMS LOST

used for the laws. Thus, the Kinetic Theory of Gases doesn’t
make use of the idea of pressure or volume at all, but only the
notion of a particle, together with its associated mass and veloc-
ity. We obtain an explanation of Boyle’s Law by deriving the
law from the principles (Newton’s laws of motion).

The idea of the gas molecules as little billiard balls flying
about inside the cylinder also illustrates the notion of a model of
a physical situation or, more precisely, a physical model as con-
trasted with a formal, or mathematical, model. No one takes seri-
ously the idea that the gas molecules really are hard little
inelastic spheres, but this turns out to be a very useful picture
upon which to let common sense feed in order to generate intui-
tions about how the physical system will act under various
circumstances. The same technique is employed in other types
of physical models, as, for instance, in the use of scale models of
cars and aircraft in wind tunnels to test for various sorts of
aerodynamic properties. In these situations, many aspects of the
real car or plane are neglected so that attention can be paid
solely to the aerodynamic properties. Similarly, in the gas exam-
ple many real properties of the gas, like its reactivity, color,
temperature, and so forth, are neglected to study its pressure-
volume relationship. Facts, laws, models, and theories — such are
the tools that the scientist uses to prospect for the gold of reality
in the mountainous doings of Nature. Figure 1.3 depicts the
interconnections between these landmarks on the terrain of
science.

Depending upon your inclination, there are several different
philosophical positions that can be taken as to whether the nug-
gets of reality that turn up in the scientist’s prospecting pan are
fool’s gold or the mother lode. In the philosopher’s game, each of
these positions is associated with a particular philosophical
point of view, or “-ism,” the most important for our purposes
being:

• Realism: Realists believe that there is an objective reality “out
there” independent of ourselves. This reality exists solely by
virtue of how the world is, and it is in principle discoverable
by application of the methods of science. I think it’s fair to
say that this is the position to which most working scientists
subscribe. They believe in the possibility of determining
whether or not a theory is indeed really true or false. Indica-

FAITH, HOPE, AND ASPERITY

Theory of
billiard
balls

Kinetic Theory

<M,V, etc.

Observations
on billiard
balls

. MODEL
(Tiny __

elastic

spheres . . .)

postulated _l
analogy

observations
on gases
(P, V, etc.)

rules of

correspondence

analogy (if any) |

between observations

FIGURE 1.3 Observations, laws, theories, and models

tive of this position is the outcome of a straw poll taken re-
cently in a small university department of physics consisting
of eleven faculty members, ten of whom claimed that what
they were describing with their symbols and equations was ob-
jective reality. As one of them remarked, “Otherwise, what’s
the use?”

• Instrumentalism: This school clings to the belief that theories
are neither true nor false, but have the status only of instru-
ments or calculating devices for predicting the results of mea-
surements. Basically, this amounts to the belief that the only
things that are genuinely real are the results of observations,
i.e., Gradgrindian Facts. A typical statement along these lines
comes from the engineer Rudolf Kalman, who remarks in the
context of mathematical model building: “[Prejudice] means
assumptions unrelated to data, independent of data; assump-
tions which cannot be (or simply are not) checked against the
data.” In light of the engineer’s hunger for any solution that
“works,” perhaps such an extreme position is acceptable in en-
gineering, but it’s hard to see how it can be defended on any
other than pragmatic grounds. As we’ll see later in the book,
the same problem arises at a much deeper level than mere
practical engineering when one passes to foundational ques-
tions of epistemology in quantum mechanics. There, too, the

(principal defense of instrumentalism is that “it works.”

• Relativism: In this increasingly popular position, truth is no
longer a relationship between a theory and an independent re-

PARADIGMS LOST

ality, but rather depends at least in part on something like the
social perspective of the person holding the theory. Thus, for a
relativist as one passes from age to age, or from society to
society, or from theory to theory, what’s true changes. In this
view it’s not what is taken to be true that changes; au contraire,
what changes is literally truth itself.

So reality is out there, in here, or what your measuring in-
struments (senses) tell you it is — take your pick! In an attempt
to tell us how to weight the odds, philosophers of science have
expended inordinate amounts of energy, thought, and heated
verbiage in pursuit of the elusive essence of the process of sci-
ence as a vehicle for unmasking the imposters on the “-ism” list.
We can summarize their Herculean task as:

THE FUNDAMENTAL QUESTION OF THE
PHILOSOPHY OF SCIENCE

Do scientists proceed as they do because there are objective rea-
sons for doing so, or do we call those procedures reasonable merely
because a certain group sanctions them?

To dig deeper into the ways science might be able to vindicate
the creed of the realists and gain a glimpse of their nirvana of
objective reality, there’s no choice but to step into the twentieth
century and look a little harder at the logical structure of sci-
ence as seen by the philosophers. While most practicing scien-
tists, not to mention laymen, find the discussion of such matters
irksome, they are inescapable and cannot be ignored in a work
such as this. Besides, as David Hawkins wisely noted, “Those
who most ignore, least escape.” So with this credo as our battle
cry, let’s briefly consider what the philosophers have to say
about the correlation between the praxis and the theoria of sci-
ence and their connection with any kind of objective reality.

RATIONALITY FOR REALISTS

If Plato’s Academy in Athens served as the geographical focal
point for Greek philosophy and its view of the world, then its
twentieth-century counterpart can only be a small seminar room
in the Mathematics Department of the University of Vienna,

FAITH, HOPE, AND ASPERITY

where a group of physicists, mathematicians, and philosophers
met every Thursday evening for several years in the 1920s and
1930s to debate the relationship between the theories of science
and objective reality. This group, christened the Vienna Circle in
1929, eventually came to what amounts to the instrumentalist
position that the only meaningful statements that can be made
are those for which we can give a definite prescription (method,
algorithm) for their verification. Thus, use of a word like “yel-
low” would be equivalent to specifying a procedure for verifying
that any particular object possessed the property of being yel-
low. In this way, the meaning or reality of “yellow” became
equivalent to the statement of the procedure for its verification.
This, in essence, forms the basis for the notorious Verification
Principle, which lay at the heart of the school of logical positiv-
ism, the term later given to the philosophy expounded by the
Vienna Circle. But to understand this blend of empiricism and
logic, it’s necessary to go back a few years and look at the work
of another Viennese philosopher of the time, Ludwig Wittgen-
stein.

WITTGENSTEIN, LOGIC, AND LANGUAGE

For ordinary men, the middle of a battlefield with bullets flying
and bombs bursting amid cries of human pain and agony is
hardly the kind of place in which to engage in contemplative
philosophical speculation. But Ludwig Wittgenstein was no or-
dinary man, and during the course of his valiant service with
the Austrian Army during World War I, he developed ideas
about the relationship of thoughts expressed in language to the
actual state of affairs in the world, ideas that were later en-
shrined in the pages of his classic work Tractatus Logico-philoso-
phicus. The basic tenet of this seminal volume, containing the
only ideas of Wittgenstein’s published during his lifetime, is
that there must be something in common between the structure
of a sentence and the structure of the fact that the sentence as-
serts. In this view, representation of the world in thought is
made possible by logic, but the propositions of logic do not in
and of themselves represent any actual state of the world. Thus,
logic was necessary but not sufficient to describe any kind of ob-
jective reality. However, for Wittgenstein logic did reveal which
states were theoretically possible, reflecting his underlying belief

PARADIGMS LOST

that reality was at least consistent — e.g., if the statement
“Water boils at 100°C at sea level” is true, then the statement
“Water does not boil at 100°C at sea level” cannot also be true.

Wittgenstein illustrated these ideas by what he called a “pic-
ture theory” of language, in which he compared logical proposi-
tions to pictures. A picture can represent some physical state
using certain types of symbols; language can do likewise but
with a different set of symbols. The picture bears some relation-
ship to the physical reality that it represents. So, for example, if
we see a human face in a photograph, the nose may appear in the
center of the face both in physical reality and in the picture.
However, if the picture is by Salvador Dali we might find the
nose appearing in some quite different location, or not at all. Of
course we might try to clarify the relationship between the pic-
ture and the object — for example, by introducing color or per-
spective— but such an attempt at clarification only gives rise to
another picture, which itself will require additional analysis. At
some stage the essence of the picture has to be understood di-
rectly, or we fall into an infinite regress.

In the picture theory of language, propositions making up the
language are thought of as analogous to a series of pictures.
Furthermore, since Wittgenstein assumes that the logical struc-
ture of language mirrors the logical structure of reality, the lan-
guage “pictures” represent possible states of the world. It
follows that linguistic statements are meaningful when they can,
in principle, be correlated with the world. Actual observation of
the world will then tell if they are true or false. To illustrate, we
can meaningfully say that “the United Nations is in New
York,” but it is meaningless to state that “is United the New in
York Nations.” Of course, different logical rules (grammars)
could be developed in which the latter statement is meaningful,
but within the context of conventional English grammar it has
no logical structure at all. So the main claim of the picture the-
ory— namely, that there must be something in common between
the logical structure of the language and the structure of the
fact that it asserts — cannot really be “said” in terms of the lan-
guage being used to make the statement; it can only be “shown.”
This conclusion gave rise to Wittgenstein’s famous metaphor in
the penultimate section of the Tractatus:

My propositions serve as elucidations in the following way: any-
one who understands me eventually recognizes them as nonsensi-

FAITH, HOPE, AND ASPERITY

cal, when he has used them — as steps — to climb up beyond them.
(He must, so to speak, throw away the ladder after he has climbed
up it.) He must transcend these propositions, and then he will see
the world aright.

So Wittgenstein’s punch line is that the sense of the relationship
between reality and its description in language cannot be ex-
pressed in language.

Thus ended Wittgenstein’s “early period” studies on the in-
terplay of logic, language, and reality. The essence of his ideas
can be summarized in the following steps:

1. There is a world that we want to describe.

2. We try to describe it in some language, scientific, mathemati-
cal, or otherwise.

3. There is a problem about whether what we say about the
world corresponds to the way the world really is.

4. We want to know the true nature of the correspondence be-
tween what we say and the way things are, but we can only
use language itself to describe that correspondence.

5. Words of a language can never express the desired correspon-
dence, and we must take recourse merely to showing it,
i.e., using the picture theory, since otherwise we would fall
into the infinite regress of descriptions of descriptions of
descriptions . . .

At Step 5 we come to one of the most famous statements in all of
philosophy, with which Wittgenstein concluded the Tractatus:
“What we cannot speak about we must pass over in silence.”

It’s easy to see how Wittgenstein’s exploration of the inter-
play of language, logic, and observation of the world would ap-
peal to the members of the Vienna Circle, with their concerns
about constructing a coherent philosophy of science from an
amalgamation of logic and empirical epistemology. And indeed
the Tractatus did serve as a point of departure for many of their
deliberations, with several members of the circle in regular con-
tact with Wittgenstein in Vienna, although Wittgenstein him-
self seems never to have participated in the Thursday night
discussions. As an ironic twist, while the Vienna Circle was busy
putting together the tenets of logical positivism using Wittgen-
stein’s work as a basis, Wittgenstein himself was in the process
of undermining the entire effort by the development of his ideas
on the rules of language.

PARADIGMS LOST

« * »

Remember those old IQ tests where some sequence of numbers
is given and you’re supposed to pick the “right” continuation of
the sequence as a demonstration of your smarts? This kind of
problem lies at the heart of what started to bother Wittgenstein
about his picture theory of language, ultimately resulting in his
repudiation of the entire idea. Consider the following simple ex-
ample. Suppose the initial sequence is (1, 2, 4, 8) and you’re
asked, what’s the “natural” or “right” continuation. On those
absurd high-school IQ and College Board tests, the examiners
would probably give full credit only if you answered with the
sequence (16, 32, 64, 128 j. Presumably this is the “correct” an-
swer because you’re supposed to recognize that each term in the
original sequence is twice as large as its predecessor. Now
there’s no doubt that this is one logically defensible reason for
guessing that the right continuation is one that extends this pat-
tern. But there can be other continuations that, depending upon
the context, would be equally logical and correct. For instance, if
the context were the high-school football stadium rather than
the examination room, then the most logical continuation might
be (1, 2, 4, 8] — (“Who do we appreciate?”). Or even in the
examination room you might think of continuing with (9, 11,
15), a pattern that reflects the jumps in the original sequence.
The point is that in the absence of context, i.e., additional infor-
mation, there’s just no such thing as a “natural” continuation
of the sequence. The reader will recognize this situation as just
another illustration of the Problem of Induction stated earlier,
and it’s just this kind of difficulty that began to trouble Witt-
genstein after the Tractatus.

Following the First World War, Wittgenstein spent time as a
high-school teacher in village schools in Austria, where it is ru-
mored he taught some of his pupils about the Liar Paradox
(“This sentence is false”). By all accounts he was very popular
with the students, but was eventually run out of the village by
their parents, most likely on account of his homosexuality and
inability to relate to the concerns of the peasant families in the
regions where he worked. In any case, during this time he began
to become dissatisfied with his picture theory of language, since
it gave no clear-cut answer to questions like “Why should we see
the principles of logic to be true, even though it’s not possible to
express the reasons in words?” (Because we can only “show”

FAITH, HOPE, AND ASPERITY

their truth, not “say” it.) Or “Is there some kind of underlying
logical structure either to the world or to our thought systems
that somehow can be held responsible for the apparent self -evi-
dence of the propositions of logic?” In other words, is there a set
of rules for organizing sense experiences that is fixed within our
brains, but that we cannot articulate even though we all follow
these rules automatically when we “see” in the same way and
when we talk to each other?

In his later work Wittgenstein considered this kind of ques-
tion, coming to the unhappy conclusion that there could be no
underlying logical structure to the world to which our minds
must adhere, or vice versa. In the final analysis, he claimed that
the propositions of logic reflect the rules of language, and these
are known to us by our use of language in everyday life and by
linguistic experience. Consequently, Wittgenstein’s solution as
to why the right continuation of the sequence [1, 2, 4, 8} is {16,
32, 64, 128} and not {“Who do we appreciate?”} is that we know
how to go on “in the same way” because we share a form of life.
Thus the continuation is dictated by sociological considerations,
and bears no contact with any kind of objective reality for num-
ber sequences. He then concluded that there are no private rules;
rules are the property of a social group. Hence, Wittgenstein
gave a “sociological” solution to the Problem of Induction by
concerning himself not with how we could be certain in principle
about the continuation, but rather with how we come to be cer-
tain about it in practice. The implication of all this for science is
that science rests upon a foundation of taken-for-granted real-
ity, a crucial aspect of the relativist school of scientific thought.
We’ll come back to this relativistic notion of scientific reality
later, but for now let’s return briefly to the Vienna Circle and
its attempts to use the early Wittgenstein to clarify the meaning
of language, thereby trying to uncover the “realness” of scien-
tific propositions about the world.

THE LOGICAL POSITIVISTS AND VERIFICATION

In his account of the evolution of knowledge, Auguste Comte
identified three stages of development: (1) the theological, in
which reality is comprehended in terms of the conflicts and crea-
tions of gods and spirits; (2) the metaphysical, in which there is
the use of abstractions and generalities; (3) the positivistic,

PARADIGMS LOST

which relies upon the quantitative description of sensory phe-
nomena. The Vienna Circle was interested in formalizing the
last stage by marrying Comte’s quantification of empirical ob-
servations and data with the logical structure of language and
its relationship to the physical world as outlined by Wittgen-
stein. The result was the philosophy of logical positivism, whose
core element was the Verification Principle discussed earlier.
For the logical positivists, there were only two sorts of state-
ments or propositions: analytic statements and those that could
be empirically verified. Only the latter had meaning, with ana-
lytic statements being either tautologies or literally meaningless.

The basic difficulty with the positivist approach is the Prob-
lem of Induction: General empirical statements just cannot be
verified. For example, if I make the empirical claim that the Sun
will rise in the east tomorrow on the grounds that it always has
risen there up to now, the Problem of Induction prevents me
from offering an empirical procedure for verifying this claim.
Consequently, according to the positivist’s creed my statement is
meaningless, and certainly not scientific. Also, the Verification
Principle had difficulties in verifying things like the wave func-
tion in quantum mechanics and, in general, failed to make a
clear-cut distinction between meaningfulness and meaningless-
ness, thus coming up empty as a criterion for meaning or real-
ity. As the source of this difficulty is the Problem of Induction,
what could be more natural than to try to get around it by the
simple expedient of rejecting the use of induction altogether?
Enter Karl Popper and the idea of falsification.

POPPER, CONJECTURES, AND REFUTATIONS

Popper, the son of a Viennese lawyer, was originally interested
in developing methods for separating scientific statements from
pseudoscience. He also took an active part in the discussions of
the Vienna Circle, whose members at first thought Popper
shared their interest in meaning, a misunderstanding that was
soon cleared up. While still a teen-ager, Popper recognized that
no amount of supporting data will ever be sufficient to confirm a
hypothesis, but all it takes to refute it is one piece of negative
evidence. So, for instance, if I hypothesize that all Ferraris are
red, no matter how many red Ferraris I see, the Problem of In-
duction will still prevent me from stating with certainty that

FAITH, HOPE, AND ASPERITY

this is the color of all Ferraris. However, all I need do is go to
the Ferrari factory in Maranello and see that there is even one
white car being built, and I can then confidently assert that my
original hypothesis is false. This chain of argument constitutes
what Popperians call the method of falsification, and forms the
heart of Popper’s view as to how science, as opposed to pseudo-
science, is to be carried out. In his own words, “The criterion of
the scientific status of a theory is its falsifiability, or refutabil-
ity, or testability.”

Popper is a realist and believes that there is an objective real-
ity out there that science can acquire increasingly accurate in-
formation about. His method is conjecture and refutation: We
make a hypothesis and then look for evidence to falsify it. For
Popper, one theory of a given situation is to be preferred to an-
other if there are more potential observations that can refute the
theory than can refute its competitor. In other words, the more
statements that could be refuted by direct observation a theory
makes, the better the theory is. The classic example is the hy-
pothesis that the Earth’s orbit around the Sun is circular, as
compared to the hypothesis (theory) that it is an ellipse with the
circular orbit as just a special case. Since there are more poten-
tial observations that will falsify, or refute, the circular hypoth-
esis, the theory that the orbit is circular would have more
empirical content for Popper. To understand clearly the distinc-
tion between Popper’s views and those of the logical positivists,
it’s instructive to examine the comparison given in Table 1.1.

While Popper seems to have banished the Problem of Induc-
tion from the philosophical banquet table, his conjectures-and-
refutations methodology is not without a few flaws of its own.
The most difficult obstacle is what is known as the Problem of
Auxiliary Hypotheses. To illustrate, let’s go back to the red Fer-
rari problem. If I happen to see a white Ferrari on the road,
thereby refuting my original contention, the “red Ferrari” hy-
pothesis can always be resurrected by adding some new back-
ground condition to the situation, such as “It wasn’t really a
Ferrari, but a Lamborghini,” or “It was a red car that had only
been painted white,” and so on. Following this line of attack,
any theory in trouble can always be saved by the introduction of
suitable auxiliary hypotheses, since it may then be claimed that
the original assertion wasn’t wrong; the error was in one of the
background assumptions.

PARADIGMS LOST

POSITIVISTS

POPPER

IDEAS

that is

DESIGNATIONS

STATEMENTS

Or TERMS

or PROPOSITIONS

or CONCEPTS

may be formulated in

Or THEORIES

WORDS

which may be

ASSERTIONS

MEANINGFUL

and their

TRUE

MEANING

may be reduced by way of

TRUTH

DEFINITIONS

to that of

DERIVATIONS

UNDEFINED

PRIMITIVE

CONCEPTS

PROPOSITIONS

The attempt to establish (rather than reduce) by these means their

MEANING

leads to an infinite regress

TRUTH

TABLE 1.1 Logical positivism versus Popper

Popper’s ideas place great emphasis upon scientific method.
He is telling scientists about how they ought to behave, neglect-
ing entirely how they actually do behave in practice. The hard
facts are that very few scientists, if any, spend much time look-
ing for data or trying to develop experiments that would falsify
their hypotheses — just the opposite, in fact. This commonplace
observation leads us into consideration of the way social conven-
tions and ideas determine what we take to be scientific truth, a
position that Popper himself ultimately came around to ac-
knowledging in connection with his original problem of distin-
guishing science from pseudoscience. He finally concluded that
if we want to know whether or not a theory is scientific, we
should look and see how it is handled by people, rather than con-

FAITH, HOPE, AND ASPERITY

sider its logical structure — a position remarkably similar to that
arrived at by Wittgenstein in his deliberations on many of the
same issues.

LAKATOS AND SCIENTIFIC RESEARCH PROGRAMS

An important way station on the road from the purely realist
position of the positivists and early Popper to the completely
relativistic stance of today’s Kuhnians, as discussed in the next
section, is the work of the Hungarian educator and philosopher
Imre Lakatos. After serving in the anti-Nazi resistance during
World War II, Lakatos became a high-ranking official in the
Ministry of Education, later fleeing to the West during the
Hungarian uprising of 1956. At this time Lakatos went to En-
gland, where he began work on his Ph.D. thesis at Cambridge on
the theme of mathematical discovery. This novel work, presented
in the form of a dialogue centering on the proof of Leonhard
Euler’s famous formula relating the number of faces, vertices,
and edges of a polyhedron, led Lakatos to a deeper interest in
the question of the “dynamics” of theories. Thus he went one
step further than Popper and the positivists by centering atten-
tion not just on the structure of scientific theories, but also upon
how they change. The vehicle for this study was what Lakatos
termed a scientific research program (SRP).

For Lakatos, an SRP is a sequence of theories in which certain
methodological rules are followed. The primary components of
an SRP are:

• The hard core — an inviolate cluster of hypotheses at the center
of the program

• The protective belt — a set of auxiliary hypotheses

• The negative heuristic — assumptions underlying the hard core
that are not to be questioned

• The positive heuristic — a set of suggestions or hints saying how
the SRP is to be altered

A good example of the kind of SRP that Lakatos had in mind is
the Ptolemaic view of the solar system, in which the Earth sits
at the center with the various planets moving about on orbits
that are described as complicated epicycles. These curves are
just the path traced out by a fixed point on, say, the rim of a
coin as you roll it along the top of a flat table. Coins of different

PARADIGMS LOST

sizes give rise to different epicycles, and Ptolemaic theory used
combinations of these curves to describe the planetary orbits.
The hard core of the Ptolemaic program is the geocentric hy-
pothesis, together with the necessity of the planetary orbits
being given by epicycles. The protective belt consists of the de-
tails of the various types of epicycles, while the positive heuris-
tic would consist of a plan for developing increasingly
sophisticated models of the planetary system. Note that this pos-
itive heuristic is not a vague, general set of principles, but a
quite specific set of procedures giving definite advice on how to
proceed, including instructions on how to handle anomalies.

On the positive side of the ledger, Lakatos’s ideas were an im-
provement over Popper’s since they acknowledged the social di-
mensions of science. In this sense they served as forerunners to
the ideas of Kuhn. Furthermore, the Lakatos vision of what con-
stitutes scientific truth had the virtue of showing that no partic-
ular research program is unambiguously to be preferred to any
other. In this way, the SRPs opened the door for the anarchical
views of Paul Feyerabend, which we’ll look at in a moment. Also
to his credit, Lakatos discerned two important facts about scien-
tific procedure: (1) scientists have sufficient faith in the hard
core that anomalies are explained away, and (2) scientists have
general ideas about how one should try to cope with anomalies
(the positive heuristic).

As to liabilities of SRPs, there are many, not the least of
which is that the choice between two SRPs for Lakatos is no
easier than the choice between two theories for Popper. The as-
sessment of which of two programs to prefer eventually comes
down to a situation analogous to having Donald Trump and
Harry Helmsley tossing pennies off the top of the World Trade
Center, the title Grand Real Estate Baron of Manhattan being
awarded to the one whose penny lands first; it’s a meaningless
game without a criterion that they can employ to see who will
reign as king of the towers. But there is no operational way for
them to decide whose penny lands first without invoking outside
agents, i.e., additional information outside the two “programs.”
Lakatos’s SRPs had other drawbacks as well.

There were great difficulties in coming to agreement as to just
what constitutes the hard core of an SRP in any specific situa-
tion. For instance, Newton’s view of planetary motion used the
inverse-square law of gravitational attraction as an inviolate hy-
pothesis, i.e., as part of the hard core of Newtonian mechanics.

FAITH, HOPE, AND ASPERITY

Yet in considering the motion of the planet Uranus, both George
Airy and Friedrich Bessel suggested modifying the inverse-
square law to account for the observations, while Ur bain Jean-
Joseph Leverrier and John Adams suggested keeping the law
and explaining the motion by the presence of a hitherto-unob-
served celestial body (which turned out to be the planet Nep-
tune). Similarly, before the Theory of Relativity was
promulgated in 1905, some suggested modifying the inverse-
square law to account for aberrations in the perihelion of the
planet Mercury. In fact, the Encyclopaedia Britannica (1910 edi-
tion) stated that the gravitational law should have the exponent
2.0000001612 instead of 2 to make things come out right! So
even in that most solid of scientific bastions, Newtonian mechan-
ics, there were heated disagreements as to what should and
should not be in the hard core. A final difficulty for Lakatos is
that the idea of the positive heuristic is hopelessly vague. This
part of the program is supposed to tell us what to do to modify
the program but, in fact, emerges during the course of the re-
search. As a result, it says nothing about what one is supposed
to do to carry out an investigation successfully.

Lakatos’s vision of the scientific enterprise is far richer than
Popper’s in that his notion of heuristics directs attention to im-
portant aspects of scientific practice not stressed by Popper at
all. Nevertheless, the difficulties with his SRPs cast aspersions
on the kinds of views of scientific “reality” that can be expected
from any such program.

So we see the various attempts by Wittgenstein & Co. to pro-
vide a solid, logical foundation, or method, for the scientific pur-
suit of knowledge all come to one bad end or another. Dare we
entertain the idea that perhaps there is no method? Well, Paul
Feyerabend not only entertains the notion, he insists upon it.

FEYERABEND: THERE AIN'T NO METHOD
In studies of scientific method, there are two principal branches:

A. Rules or techniques to use in the discovery of theories

B. Rules for the objective evaluation of rival theories

The Vienna Circle claimed that only B was the legitimate prov-
ince of the philosophy of science; Paul Feyerabend denies that
there is any valid distinction between the two.

In Against Method, his famous manifesto for scientific
anarchy, Feyerabend states his basic theme in the following way:

PARADIGMS LOST

“No set of rules can ever be found to guide the scientist in his
choice of theories, and to imagine there is such a set is to impede
progress. The only principle that does not impede progress is
anything goes [italics added].” Feyerabend is claiming that there
is no such thing as a scientific method. His argument is that
science is just one tradition among many, and is privileged nei-
ther in terms of methods nor in terms of results. He goes on to
advocate removing science from its pedestal and trying to create
a society in which all traditions have equal access to power and
education. Among the traditions he suggests giving equal weight
with science are astrology, witchcraft, mysticism, and folk medi-
cine! If this all sounds like the grumblings of a failed scientist
to you, it’s perhaps worth noting that Feyerabend did at one
time study physics and astronomy.

Feyerabend was also active in the Berkeley Free Speech
Movement, and became interested in the so-called alternative so-
ciety ideas bandied about in the 1960s. But he eventually re-
deems himself by confessing that he doesn’t have the seriousness
of purpose of a true anarchist and would like to be remembered
as a “flippant Dadaist.”

The central thesis of incommensurability of theories brought
out in such stark fashion by Feyerabend takes us from the ideas
of realism and the work of Wittgenstein and the Vienna Circle
clear across town to relativism and the offbeat ideas of Feyera-
bend. Despite their shade of lunacy, the visions of Feyerabend
contain just enough good sense to suggest there’s something
worthwhile lurking at their core. This kernel of sense hiding in
the flamboyant noise is the notion that there are many methods
and ways of coming to scientific truth, and what is taken to be
true at any moment is more a matter of social convention in the
scientific community than it is a product of logical methods and
procedures. Recognition of this startling fact constitutes the
theme song for Thomas Kuhn, whose ideas about paradigms in
science lie at the heart of what is by far the most talked-about
view of the scientific enterprise in the second half of this cen-
tury.

BUDDY, CAN YOU PARADIGM?

Julian Bigelow, an electrical engineer who helped John von Neu-
mann build the Johnniac computer at the Institute for Ad-

FAITH, HOPE, AND ASPERITY

vanced Study in Princeton in the early 1950s, tells a story about
how when he drove down from Cambridge, Massachusetts, to be
interviewed by von Neumann for the job, he met with the great
man at his home in Princeton. As the story goes, there was a
large dog romping on the lawn, and as von Neumann opened the
door to let Bigelow in, the dog ran into the house and started
running from room to room, sniffing everything in sight in the
manner commonly practiced by dogs everywhere. Busy in their
discussion, neither von Neumann nor Bigelow paid much atten-
tion to these canine antics for quite awhile, but finally von Neu-
mann’s curiosity overcame his courtly Central European
manners and he asked Bigelow if he always traveled with his
dog. Bigelow replied, “It’s not my dog. I thought it was yours.”
Such are the presuppositions that pervade every aspect of
human activity, science (and scientists) being no exception. And
it’s exactly these kinds of presuppositions that constitute the
nucleus of the idea underpinning Thomas Kuhn’s notion of a
scientific paradigm.

In 1947 Kuhn, a young professor at Harvard, was asked to
organize a set of lectures on the origins of seventeenth-century
mechanics. As preparation, he began tracing the subject back to
its roots in Aristotle’s Physics, being struck time and again by
the total and complete wrongheadedness of Aristotle’s ideas. As
noted earlier, Aristotle held that all matter was composed of
spirit, form, and qualities, the qualities being air, earth, fire,
and water. Kuhn wondered how such a brilliant and deep
thinker, a man who had single-handedly invented the deductive
method, could have been so flatly wrong about so many things
involving the nature of the physical world. Then, as Kuhn re-
counts it, one hot summer day the answer came to him in a flash
while he was poring over ancient texts in the library: Look at the
universe through Aristotle’s eyes! Instead of trying to squeeze
Aristotle’s view of things into a modern framework of atoms,
molecules, quantum levels, and so forth, put yourself in Aris-
totle’s position, give yourself the prevailing world view of Aris-
totle’s time, and lo and behold, all will be light. For instance, if
you adopt Aristotle’s world view, one of the presuppositions is
that every body seeks the location where by its nature it belongs.
With this presumption, what could be more natural than to
think of material bodies as having spirits, so that “heavenly”
bodies of airlike quality rise, while the spirit of “earthly” bodies
causes them to fall?

PARADIGMS LOST

FIGURE 1.4 Two visual gestalts or “paradigms ”

This stroke of inspiration resulted in Kuhn’s developing the
idea that every scientist works within a distinctive paradigm, a
kind of intellectual gestalt that colors the way Nature is per-
ceived. The situation is vaguely analogous to the picture in Fig-
ure 1.4, where one way of looking shows what appears to be two
men face to face in profile, while another way shows a flower
vase.

According to Kuhn’s thesis as presented in his enormously in-
fluential 1962 book The Structure of Scientific Revolutions, scien-
tists, just like the rest of humanity, carry out their day-to-day
affairs within a framework of presuppositions about what con-
stitutes a problem, a solution, and a method. Such a background
of shared assumptions makes up a paradigm, and at any given
time a particular scientific community will have a prevailing
paradigm that shapes and directs work in the field. Since people
become so attached to their paradigms, Kuhn claims that scien-
tific revolutions involve bloodshed on the same order of magni-
tude as that commonly seen in political revolutions, the only
difference being that the blood is now intellectual rather than
liquid — but no less real! In both cases the argument is that the
underlying issues are not rational but emotional, and are settled
not by logic, syllogisms, and appeals to reason, but by irrational
factors like group affiliation and majority or “mob” rule. As
Kuhn states it: “There is no standard higher than the assent of

FAITH, HOPE, AND ASPERITY

the relevant community. The transfer of allegiance from one
paradigm to another is a conversion experience that cannot be
forced.” With these ideas in mind, just what constitutes a para-
digm anyway, at least as that term is used by Kuhn? The answer
is not easy, and Kuhn has come in for plenty of criticism for the
vagueness of the notion. But the basic concept can be made clear
by the following map-making analogy.

Let’s imagine scientific knowledge of the world as being the
terra incognita of the ancient geographers and map makers. In
this context, a paradigm can be thought of as a crude sort of
map in which territories are outlined, but not too accurately,
with only major landmarks like large rivers, prominent moun-
tains, and the like appearing. From time to time, explorers ven-
ture into this ill-defined territory and come back with accounts
of native villages, desert regions, minor rivers, and so on, which
are then dutifully entered on the map. Often such new informa-
tion is inconsistent with what was reported from earlier expedi-
tions, so it’s periodically necessary to redraw the map totally in
accordance with the current best estimate of how things stand in
the unknown territory. Furthermore, there is not just one map
maker but many, each with a different set of sources and data on
the lie of the land. As a result there are a number of competing
maps of the same region, and the adventurous explorer has to
make a choice of which map he will believe before embarking
upon an expedition to the “New World.” Generally, the explorer
will choose the old, reliable firm of map makers, at least until
gossip and reports from the Explorers Society show too many
discrepancies between the standard maps and what has actually
been observed. As these discrepancies accumulate, eventually the
explorers shift their allegiance to a new firm of map makers
whose pictures of the territory seem more in line with the re-
ports of the returning adventurers.

This exploration fable gives a fair picture of the birth and
death of a scientific paradigm. Kuhn realized that revolutionary
changes in science overturning old theories are not in fact the
normal process of science, nor do theories start small and grow
more and more general as claimed by Bacon, nor can they ever
be axiomatized as asserted by Newton. Rather, for most scien-
tists major paradigms are like a pair of spectacles that they put
on in order to solve puzzles. Occasionally a paradigm shift takes

PARADIGMS LOST

place when the spectacles get smashed, and they then put on a
new pair that transforms everything into new shapes, sizes, and
colors. Once this shift takes place, a new generation of scientists
is brought up wearing the new glasses and accepting the new
vision of “truth.” Through these new glasses, scientists see a
whole new set of puzzles to be solved in the process of carrying
out what Kuhn called normal science.

The paradigms have great practical value for the scientist just
as maps have value for the explorer: Without them no one would
know where to look or how to plan an experiment (expedition)
and collect data. This observation brings out the crucial point
that there is no such thing as an “empirical” observation or
fact; we always see by interpretation, and the interpretation we
use is given by the prevailing paradigm of the moment. In other
words, the observations and experiments of science are made on
the basis of theories and hypotheses contained within the pre-
vailing paradigm. As Einstein put it, “The theory [read para-
digm] tells you what you can observe.” According to Kuhn’s
paradigmatic view of scientific activity, the job of normal sci-
ence is to fill in the gaps in the map given by the current para-
digm, and it’s only seldom, and with great difficulty, that the
map gets redrawn when the normal scientists (explorers) turn
up so much data not fitting into the old map that the map begins
to collapse into a morass of inconsistencies. But what happens
during these times of paradigm crisis?

Imagine we are at the initial stages of such a crisis, where the
old paradigm can’t account for certain anomalies, strange obser-
vations, and the like. Two new theories emerge, which offer dif-
ferent explanations for these aberrations. These theories
represent different maps or sets of spectacles, i.e., different reali-
ties. After a period of competition, one of these theories begins
to gain the acceptance of the scientific community. The reasons
may not be objective at all, but may revolve about matters like
simplicity, elegance, the social position of the theory’s adherents,
government science policies, and so forth. This support leads to
experiments that then “corroborate” the theory, and the more
evidence that accumulates, the more supporters the theory gath-
ers, especially among the young Turks in the scientific commu-
nity. Soon “reality” begins to take on the look of the new
theory, and scientists universally begin to see and test for cer-
tain features of this reality and ignore others.

But what if the community had given its initial support to the

FAITH, HOPE, AND ASPERITY

other, competing theory? According to Kuhn, in that event “re-
ality” would have taken a quite different turn, and the scientific
view of the world would have been seen through that pair of
spectacles rather than the first. This means that there is no such
thing as scientific “progress,” at least not in the sense that one
paradigm builds upon its predecessor. Rather, the new paradigm
turns in an entirely different direction, and as much knowledge
is lost with the abandonment of the old paradigm as is gained
from the new. Now we “know” a different universe.

If Kuhn’s thesis is true, then it also destroys one of the main
pillars of the scientific method, since the whole idea of a scien-
tific experiment rests upon the assumption that the observer can
be essentially separate from the experimental apparatus that
tests the theory. Kuhn contends that the observer, his theory,
and his equipment are all essentially an expression of a point of
view, and the results of the experimental test must be an expres-
sion of that point of view as well. This position effectively as-
serts that science is not objective, but at the same time we know
that science is not totally subjective either, since paradigms are
eventually overthrown. So we’re back to consideration of the
central question: What is the relationship of the scientist to the
universe he observes?

The most revolutionary aspect of Kuhn’s claims is that they
entirely omit things like knowledge, truth, and external reality.
In fact, Kuhn states that in science truth is an entirely optional
and gratuitous concept. As he puts it, “Does it really help to
imagine that there is some one full, objective, true account of
nature and that the proper measure of scientific achievement is
the extent to which it brings us closer to that ultimate goal?” I
think most practicing scientists would say that yes, such a belief
helps a hell of a lot! But apparently Kuhn doesn’t think so, since
he says that there’s no way for science to get hold of the “truth”
anyway, so you can’t measure scientific progress as getting
closer to the way things are in themselves. Returning to the
map-making analogy, Kuhn’s claim is tantamount to the belief
that not only are there many map makers, each emphasizing dif-
ferent aspects of the territory, but that it is in principle impossi-
ble ever to produce a complete map of the entire region. So you
can’t judge a map by how close it comes to this ideal Platonic
map, since such a map is literally undrawable. In some ways this
line of argument is reminiscent of Wittgenstein’s claim that lan-

PARADIGMS LOST

guage cannot describe the intrinsic logical structure of the
world.

Just like the revolutions they describe, Kuhn’s arguments
were met with fierce opposition from the philosophical commu-
nity, although he was a minor saint to humanists since he
seemed to be putting the human being back into the scientific
enterprise. One of Kuhn’s sharpest critics has been the philoso-
pher Dudley Shapere, who complained that Kuhn was a relati-
vist denying the objectivity and rationality of science. Shapere
felt that science according to Kuhn is nothing more than a series
of fads dressed up to look presentable, and offered the coun-
terargument that even though we may be wearing rose-colored
glasses, there’s still a lot that shines through unaffected. The col-
ors may be skewed, but other qualities like shape, size, and tex-
ture come through loud and clear. In short, the glasses may
distort our view of reality but they don’t create it — a staunch
realist position.

Another criticism of Kuhn’s ideas is that he places too little
emphasis upon the social determinants of scientific revolutions.
On the one hand, Kuhn argues that a paradigm shift takes place
when there’s an accumulation of anomalies; on the other hand,
he says an anomaly can be ignored to preserve the paradigm.
Question: At what point does a mass of discrepancies become ir-
ritating enough to bring about a paradigm shift1? Kuhn offers
little help in addressing this dilemma.

While Kuhn denies the label of an “irrationalist,” he does as-
sert that there are no methods or methodological rules for creat-
ing or evaluating scientific theories. His argument is that only
propagandizing plays a role in changing allegiances from one
paradigm to another. What makes reasons for theory change
“good” is that they are generally accepted by the community,
and if you want to be a member of that community it behooves
you to operate within the framework of this system of reasons.
As an immediate consequence, we find Kuhn’s statement that
rival paradigms cannot really be compared, although he does
offer what we might term a Fivefold Way for characterizing the
features of a good theory. Kuhn’s way consists of the following
points stating that a good theory must be

• Accurate: Consequences of the theory should be in agreement

with experiment.

FAITH, HOPE, AND ASPERITY

• Consistent: The theory should contain no internal contradic-
tions and, moreover, it should be consistent with currently ac-
cepted theories applicable to related aspects of Nature.

• Broad: The scope of the theory’s consequences should extend
beyond the particular observations, laws, or subtheories that it
was created to explain.

• Simple: It should bring order to phenomena that without it
would be individually isolated.

• Fruitful: The theory should disclose new phenomena or previ-
ously unobserved relationships.

Kuhn’s claim is that these criteria offer the shared basis for the-
ory choice, but that there is no possible way of giving a justifi-
cation for this selection of criteria.

To compare Kuhn with Feyerabend, Kuhn says there are
rules (the Fivefold Way) for theory choice, but their application
may be problematic and they cannot be given objective justifica-
tion. Feyerabend says there are no rules whatsoever but, like
Kuhn, rests much of his case on the existence of incommensura-
ble theories.

We can also compare Kuhn with Popper and Lakatos by not-
ing that, roughly speaking,

Paradigm = Hard core + Positive heuristic

enabling us to connect Lakatos’s SRPs to the notion of a para-
digm. As far as Popper is concerned, his central themes of con-
jecture, test, refutation, are also present in Kuhn’s world, but
only during the course of practicing normal science. Popper’s
contention that there is no rationale for the introduction of new
conjectures in science, but only for the exposure of such conjec-
tures to falsifying tests, is basically similar to Kuhn’s claim that
there is no rationale for the introduction of a new paradigm, but
only for the attempt to “articulate” the paradigm and make it
deal successfully with anomalies. The point of divergence be-
tween Kuhn and Popper arises when it comes time to shift from
one paradigm to another. Popper believes this can and should
(and is) done rationally, logically, and with little fuss; Kuhn
says this method may be fine in the abstract, but real science
just doesn’t work that way.

With Kuhn we have come to the end of the line as far as con-
temporary views on the ways science operates both to form and

PARADIGMS LOST

to validate its view of the world. Since the path from Wittgen-
stein to Kuhn has been a complicated one filled with lots of
switchbacks and strange meanderings, in the next section I’ll try
to summarize the competing positions as well as briefly reexam-
ine our original question: How real is scientific reality?

PHILOSOPHICALLY SPEAKING

When embarking upon this whirlwind tour of twentieth-century
philosophy of science, our point of departure was to explore the
two basic issues: What is the connection between scientific theo-
ries (language) and objective reality, and does science have any
special sort of procedures or methods for either generating new
theories or evaluating competing ones? Note again here the im-
portant point that when we use the term method in this setting,
we’re referring to a method for generating theories and not to
the more common concept of the “scientific method” as con-
stituting the potentially infinite sequence hypothesis -*■ experi-
ment -» hypothesis . . . These questions led us to divide beliefs
on the nature of reality into three categories:

• Realism = Objective reality exists.

• Instrumentalism = Reality is the readings noted on measuring

instruments.

• Relativism = Reality is what the community says it is.

We also saw that beliefs as to whether or not there’s method
in the madness of science determine one’s position as a rational-
ist or an irrationalist, with rationalists believing in method, ir-
rationalists not. The various philosophers and philosophical
schools took differing views on these matters, and to expound
them occupied a lot more time and space than I’d intended, but
necessarily so. Consequently, before going on to consider what
the practicing scientists themselves, as well as competing ideolo-
gies, have to say about these matters, I have tried to summarize
the story so far in Table 1.2. As the table shows, the overwhelm-
ing conclusion of the philosophers is that, as Einstein said, “it’s
all relative.” But we saw earlier that ten out of eleven everyday
physicists supported the idea of an objective reality “out there”
that their equations were describing. To address this paradox,
let’s quickly hear from the laboratory instead of the ivory tower

FAITH, HOPE, AND ASPERITY

SCHOOL

REALITY BELIEF

METHOD

ARGUMENT

Wittgenstein I

realism

rationalist

picture language

Wittgenstein II

relativism

irrationalist

language rules

logical positivists

instrumentalism

rationalist

verification

principle

Popper

realism

rationalist

falsification

Lakatos

relativism

rationalist

SRPs

Feyerabend

relativism

irrationalist

“anything goes”

Kuhn

relativism

rationalist

paradigms

TABLE 1.2 The battle of the philosopher kings

and listen to what the players rather than the Monday morning
quarterbacks have to say about the whole business.

In 1979 the Institute for Advanced Study in Princeton held a
celebration to honor the one hundredth anniversary of the birth
of Einstein, the institute’s first and most celebrated resident ge-
nius. To plan for this celebration, a committee was formed at the
institute to arrange a program and invite scholars from around
the world to participate. Just as Caesar divided all Gaul into
three parts, the IAS committee decided to organize the Einstein
centennial similarly, focusing on Einstein’s science, the histori-
cal genesis of his ideas, and, finally, the philosophical impact of
his work. As Freeman Dyson tells it, the committee solicited
names and put together lists of scholars who could be invited in
each of the three areas. The committee was personally ac-
quainted with almost everyone on the list of scientists. As to the
historians, the committee didn’t know them personally but at
least had heard of most of them and knew of their work. But
when it came to the philosophers of science, Dyson remarks that
the committee was not only unfamiliar with them personally, but
had never even heard the names of most of them! More than any
abstract argument could ever hope to show, this little episode
conveys the level of contact between the activities of the working
scientist and the arguments of the philosopher: It is exactly
zero! In Dyson’s words, “There’s a whole culture of philosophy
out there somewhere with which we have no contacts at all. . . .
there’s really little contact between what we call science and
what these philosophers of science are doing — whatever that is.”

Dyson’s observation serves to unravel the contradiction noted

PARADIGMS LOST

a moment ago between the beliefs of scientists and those of phi-
losophers. As far as most practicing scientists are concerned,
there’s nothing more dangerous than a philosopher in the grip of
a theory. In fact, there appears to be something of an unre-
quited love affair between the scientists and philosophers, in
which the scientists by and large spend their days ignoring the
attempts by the philosophers to press their attentions upon
them. As an indicator of the state of affairs, the physicist Mur-
ray Gell-Mann at all times carries with him a doctor’s prescrip-
tion forbidding him to argue with philosophers on the grounds
that it could be dangerous to his health!

So we come to the perhaps not so surprising conclusion that if
you want to know about how scientists really think and work,
you’ll get no help from a philosopher of science. However, if
your concerns go beyond what scientists do and encompass the
broader issues of the significance of what they do and its relation-
ship to other knowledge-generating mechanisms, then, as noted
before, a consideration of matters philosophical is unavoidable.
Most of our stories in this volume center upon what scientists
are really doing, but in each one of them there is a strong under-
current of philosophical presupposition conditioning the inter-
pretation of the results. The reader should try to keep these
deeper issues in mind as we go along, as a guide to evaluating
the myriad competing arguments.

While philosophical factors probably are honored more in the
breach than in the practice of science, sociological pressures are
another matter. Science is not yet done by impersonal, unin-
volved machines, but by real, live, thinking and feeling human
beings, and it’s impossible for this fact not to have some impact
upon the way science proceeds to its conclusions about the way
the universe functions. Let’s take a few pages to consider the
sociology of science rather than its philosophy, as another ave-
nue to walk down on our way to learning about the way science
comes to what it sees as “truth.”

A TALE OF TWO SUICIDES

Ludwig Boltzmann and Paul Kammerer were both professors at
the University of Vienna in the early part of this century; they
were both popular with their students and held in great esteem

FAITH, HOPE, AND ASPERITY

by their colleagues; and they both committed suicide. While per-
haps extreme in the outcome, these two cases serve as examples
of one aspect of the way scientific truth is determined at least as
much by the social climate of the times as by the dictates of
reason and logic alone.

Boltzmann, a physicist, is perhaps best remembered for his
work in thermodynamics and the connections he discovered be-
tween the theory of heat and the more general issues of random-
ness and order. He is today credited with having introduced the
notion of entropy as a measure of the disorder present in a col-
lection of objects of any sort, an idea that later served as the
basis for the theory of information, which turned out to be so
crucial to the development of modern communications technol-
ogy. In fact the formula S = k log W, expressing the entropy S
as being proportional to the logarithm of W, the number of pos-
sible states that a system can assume, is engraved on Boltz-
mann’s tombstone in Vienna’s Zentralfriedhof, a fitting
memorial to the importance of this fundamental idea. In this
expression, the constant of proportionality k is even today
termed Boltzmann’s constant in recognition of this magnificent
achievement. But at the time he was carrying out this pioneer-
ing work, the achievement was anything but magnificent, at least
if one was listening to the leading scientists of the day.

Boltzmann’s problem was that his theory of heat involved an
assemblage of atoms moving according to the usual rules of
Newtonian mechanics. He used this concept of an atom as a par-
ticle of matter to construct his theory of heat as a statistical
property emerging out of the overall motion of these atoms.
Note that this idea was put forth around the turn of the cen-
tury, several years before the work of Ernest Rutherford, J. J.
Thomson, and Niels Bohr gave the concept of an atom its mod-
ern birth. As a result of his atomistic speculations, Boltzmann
came into heated conflict with several of the giants of the scien-
tific community, most notably his Viennese colleague Ernst
Mach and the German physical chemist Wilhelm Ostwald, who
argued forcefully against the idea of the atom. Ostwald, in par-
ticular, preferred a theory of heat based upon the notion of en-
ergy rather than matter. Depressed by the acrimony of this
opposition, as well as his failing eyesight and what he thought of
as the decline of his mental faculties, Boltzmann took his life in
Duino, Italy, on September 5, 1906.

PARADIGMS LOST

Tragically, Boltzmann’s suicide took place almost cotermi-
nously with the work by Thomson and Rutherford in Britain
that would lead to a complete vindication of his ideas. So here
we have a textbook illustration of how the social climate of the
scientific community, as well as the influence of two great men,
acted to delay introduction of what ended up being a major con-
tribution to our way of thinking about the way the world works.
Now let’s move the clock forward almost exactly twenty years
and examine the case of another Viennese professor as illustra-
tion of how these same social forces can work to rid science of
equally controversial, but this time erroneous, ideas.

Paul Kammerer was a professor of biology at the University
of Vienna in the 1920s. Accounts credit him with an almost mag-
ical skill at breeding amphibians and other types of animals.
They also note that he was an ardent socialist and crusader for
the political causes of what today we would term the liberal left.
Given this combination of scientific and political leanings, it’s
perhaps not surprising that Kammerer supported the idea that
acquired characteristics can be pass on to offspring, i.e., La-
marckian inheritance. For ideologues bent upon improving the
human race, the idea that behavioral traits like learning, altru-
ism, and the like can be acquired holds great appeal. So it was
for Kammerer, too, and he set out to prove the idea with his now
infamous experiments on the midwife toads.

Generally these toads breed on land, with the male lacking the
so-called nuptial pads of the male members of other species of
toads that breed in the water. These pads are rough patches on
the hands of the male that he uses to grab on to the back of the
slippery female during the course of mating in water. Kam-
merer’s experiment involved forcing the midwife toad to breed
in water for several generations, his claimed results being that
such toads then developed the nuptial pads characteristic of
their naturally water-breeding cousins. The supporters of Kam-
merer focused upon this experiment as clear-cut evidence for
Lamarckism; opponents remained highly doubtful and requested
a closer look at the evidence.

These experiments with the midwife toad came under heavy
attack from naturalists in both Europe and America, especially
William Bateson in England and Kingsley Noble in New York.
On a visit to Vienna in 1923, Bateson saw Kammerer ’s last re-

FAITH, HOPE, AND ASPERITY

maining specimen of a midwife toad with nuptial pads and later
asked to reexamine it in his own lab. Kammerer replied that it
could not be sent from Vienna. At the same time, Noble was hav-
ing doubts about some of the particulars of the physical struc-
ture of Kammerer’s claimed nuptial pads, and visited Vienna in
1926 to examine the last specimen personally. His results, pub-
lished later that year in Nature, claimed that the so-called pads
were nothing more than black markings made with India ink.

At the time of Noble’s report, Kammerer was preparing to
leave Vienna for a position at Moscow University as head of a
new laboratory in Lamarckian biology. Noble’s Nature article
appeared on August 7, 1926. In a letter of September 22 to the
Soviet Academy of Sciences, Kammerer wrote that he had exam-
ined Noble’s claims and found them to be totally accurate. He
went on to protest his ignorance of how the inking had been
done, but acknowledged that his experimental conclusions about
Lamarckism were baseless. After withdrawing from the post in
Moscow, the letter concluded with the poignant statement “I am
not in a position to endure this wrecking of my life’s work, and
I hope that I shall gather together enough courage and strength
to put an end of my wrecked life tomorrow.” And, in fact, dur-
ing a walk in the Wienerwald the next day, Kammerer shot him-
self in the head. This was another extreme example of scientific
peer-group pressure and its sometimes tragic effect upon the
lives of scientists deviating from the group norms. Only this
time the pressure acted to discredit wrong results rather than to
suppress correct ones.

The tales of these two Viennese professors serve to under-
score the sometimes dramatic influence that the social compo-
nent of science plays in establishing what we take to be the
scientific “truth” of the moment. These social factors operate
within the scientific community itself as well as in the outside
world, shaping not only the way scientific activity is carried
out but also the manner in which certain ideas, like Boltz-
mann’s, are buried while others thrive. One of the pioneers in
studying these social determinants, at least inside science it-
self, is the sociologist of science Robert K. Merton, who in
1942 identified a small set of what he termed norms character-
izing the scientific enterprise. Roughly speaking, in modem
terms we can give Merton’s norms as:

PARADIGMS LOST

• Originality: Scientific results should always be original, i.e.,
novel. Studies that add nothing new to what is already known
are not part of science.

• Detachment: Scientists undertake their work with no motive
other than the advancement of knowledge. They should have
no personal axes to grind insofar as the results of their work
go, and they should have no psychological commitment to a
particular point of view. The impersonal style of most scien-
tific communications is a direct consequence of this norm.

• Universality: Claims and arguments should be given weight ac-
cording to their intrinsic merits alone, and should not depend
upon religious, social, ethnic, or personal factors surrounding
the individuals who make them. In short, there are no privi-
leged sources of scientific knowledge.

• Skepticism: No scientific statements of fact should be taken on
faith. All claims should be carefully scrutinized for invalid ar-
guments and errors of fact, and any such mistakes should be
made public immediately. To put it simply, scientists should
trust no one, at least not when it comes to claims of scientific
truth.

• Public accessibility: All scientific knowledge should be freely
available to anyone. Thus, results of research are not the pri-
vate property of the scientist, but are public goods that should
be transmitted immediately to the community of science. This
norm lies at the heart of debates as to whether or not engaging
in classified military research is scientifically ethical.

Anyone involved with the way scientific practice actually
works will immediately recognize that these prescriptions are vi-
olated every day of the week in both trivial and not so trivial
ways, serving the same role in science that general laws serve for
society at large. There’s nothing particularly disturbing about
this gap between theory and practice, just as the fact that
human beings jaywalk, rob banks, and drive their cars too fast
is not really news either. What is disturbing, to some anyway, is
what appears to be an increasing incidence of such violations of
the spirit of science, at least as it’s embodied in these norms.
Such an increased pace of corner cutting in science seems espe-
cially evident in the last decade or so, certainly aided and abet-
ted by science’s Faustian bargain with government funding
agencies. Nevertheless, the Mertonian norms are still the ethos to

FAITH, HOPE, AND ASPERITY

which the community of scientists subscribes, and form the
heart of the code by which the behavior of most scientists is
judged by their peers. And in exactly this way the norms make
their contribution to the way scientists think, hence to what they
ultimately come to accept as the way things are. But these fac-
tors working inside the scientific community are not the only so-
cial components influencing the work of science. Of equal
importance are the forces affecting science from the outside, es-
pecially in today’s mass-media-saturated and cash-hungry
world.

In his 1971 State of the Union address, President Richard M.
Nixon declared that the time had come for the country to wage
war on cancer, with the “same kind of concentrated effort that
split the atom and took man to the Moon. . . This pronounce-
ment led to an avalanche of money pouring into the nation’s can-
cer research laboratories, and resulted not only in a war on
cancer but also in a war among the various research establish-
ments for a generous hunk of the federal government’s cancer
war chest. One of the foot soldiers in both of these conflicts was
William T. Summerlin, a young skin specialist at the prestigious
Sloan-Kettering Institute for Cancer Research in New York
City.

Amid the high-pressure political climate surrounding cancer
research and the feverish hustling and grantsmanship, in March
1973 Summerlin applied for a five-year federal research grant
from the American Cancer Society to pursue his special interest
in skin grafts and immunology. In particular, Summerlin felt
that he was on the track of developing procedures whereby skin
treated by his technique could be transplanted without rejection.
Thinking that a little favorable publicity never hurt the case of
a relatively obscure, but ambitious, young researcher, Summer-
lin presented an outline of his work in progress at a science writ-
ers’ convention. The results were predictable: a three-column
headline the next day in The New York Times declaring lab dis-
covery may aid transplants. Summerlin was on his way, or so it
seemed.

During the course of the next year, while Summerlin traveled
the country presenting seminars and lectures on his work, col-
leagues were finding it increasingly difficult to confirm his re-
sults by independent experiments. In fact, even workers in

PARADIGMS LOST

Summerlin’s own laboratory at Sloan-Kettering were unable to
reproduce the claimed properties of the specially treated “Sum-
merlin skin,” leading to a showdown between Summerlin and
Sloan-Kettering Director Dr. Robert A. Good in March 1974. On
his way to this fateful meeting, Summerlin pulled out a black
felt-tip pen and hurriedly inked in some dark patches on the
white mice he was bringing as evidence for his claims. At the
time Good didn’t notice the Summerlin embellishments, and it
was only when the mice were returned to the lab assistant that
Summerlin’s “help” was discovered. The assistant immediately
reported the matter to his boss, at which point Summerlin was
instantly suspended. While he denyed any wrongdoing, assert-
ing that he had inked in the skin grafts on the mice only to make
them more easily identifiable, Summerlin’s credibility was shat-
tered by the incident, along with the credibility of his supposed
technique for skin grafts.

Interestingly enough, the Summerlin episode bears some
strange similarities to that of Kammerer and the midwife toads,
although without the same tragic suicidal ending. The point in
raising these cases here is not so much the issue of whether or
not Kammerer or Summerlin was really guilty of fraud, but
rather to illustrate the degree to which forces outside the world
of science, in this case the federal research-funding establish-
ment and the public at large, contribute to creating a climate
that can drive scientists to manufacture and/or artificially en-
hance what they claim are “the facts.” And money is not the
only such pressure. Political considerations, especially those in-
volving what is often termed “human nature,” can and do play a
dramatic role in influencing what’s scientifically “right.” A good
illustration of this kind of effect was the controversy over social
Darwinism in the first half of the century, a debate about which
we shall have much more to say later when we consider its mod-
em incarnation: the Sociobiology Problem. In this context, it
may even be safe to say that the real issue is the conflict between
the norms of science, as exemplified by Merton’s list, and the
“norms” of politics as encoded in the ideologies of certain politi-
cal movements (in the case of sociobiology, Marxism).

The foregoing stories barely scratch the surface of the many
ways in which sociological considerations shape what science
thinks of as being true, with many far more detailed accounts
noted under “To Dig Deeper” in this volume. For our purposes

FAITH, HOPE, AND ASPERITY

here, the main consideration is the manner in which these social
factors influence the way science validates its claims and comes
to a consensus on a given issue. The heart of the difficulty is that
knowledge is underdetermined. Thus, there are always many
different theories, each of which can give a plausible account of
the available facts. So how are we to choose one and let the oth-
ers gof The basic problem is encapsulated in the remark of the
philosopher Willard Van Orman Quine, who noted that “any
statement can be held true, come what may, if we make drastic
enough adjustments elsewhere in the system.” One natural place
to make these drastic adjustments is in the cultural background
to the problem, thereby creating a climate in which only one or
at most a few of the contending theories can survive. Again, we
will see ample evidence of this kind of “cultural imperialism” in
the raging sociobiology debate covered in Chapter Three.

As to arrival at a consensus, the key factor is the Mertonian
norm relating to the public character of scientific knowledge.
The rule that scientific information is communicated explicitly
and unambiguously influences both the form and the content of
knowledge that is labeled “scientific.” For example, this norm
goes a long way toward accounting for why experimental verifi-
cation involving neutral instrumentation occupies such a hal-
lowed position in science, as well as the great value attached to
quantitative observation and expression of results in mathemati-
cal form. All of these features contribute to the public accessibil-
ity of the information and the reproducibility of the results, at
least in principle. One need only consider other fields like litera-
ture or the arts, where such a norm is not the norm, to see some
of the ways in which scientific knowledge differs in significant
ways from these other forms of reality representation.

Since we’ll see many concrete instances of these sociological
factors entering into the stories that follow, there’s no need to
belabor the point here in the abstract. For now, it’s of somewhat
more interest to look at some of the knowledge-generating de-
vices that make some pretense to a degree of scientific character,
in their goals if not their methods. With the above ideas as prel-
ude, the reader should be in a better position to distinguish those
groups doing what we would now term science from those prac-
ticing at the fringe.

We began this chapter with the dual stories of Jocelyn Bell
and Immanuel Velikovsky, noting their positions at opposite

PARADIGMS LOST

ends of the spectrum of what’s currently held to be “good sci-
ence.” We are finally in a position to give the long answer to the
question posed earlier about why Velikovsky’s work has been
relegated to the dustbin of pseudoscience, while Bell’s was re-
warded with the Nobel Prize for physics (although not to her).

ON THE FRINGE OR AT THE CUTTING EDGE f

As editor of a scientific journal, I’m regularly faced with the
unpleasant task of telling potential contributors that their pa-
pers are not suitable for publication. Generally the reasons are
the usual ones: trivial or nonexistent results, poor writing, work
outside the scope of the journal, and so on. However, occasion-
ally I get a paper that I don’t even bother to send out for the
customary refereeing process, rejecting it out of hand. Such pa-
pers are the bane of the editor of almost every scientific publica-
tion, and every editor soon becomes sensitized to their telltale
aroma of nonsense masquerading as science. Since my own jour-
nal is devoted to mathematics, papers of this sort tend to involve
such well-known impossibilities as squaring the circle, trisecting
an angle, and doubling the cube, although they occasionally ad-
dress famous outstanding problems like Fermat’s Last Theorem
or the Riemann Hypothesis (in which case I’m compelled to look
at them seriously, even though there’s not yet been one that was
correct). Luckily for me, mathematics is an area where it’s dif-
ficult to try to dress up such pseudoscience in respectable clothes
and not have it show. Certainly my colleagues in biology, medi-
cine, and the social sciences must have it much worse in this re-
gard. But just what is it about this kind of paper that
immediately stamps it as pseudoscience to the trained (and jaun-
diced) scientific eye? To answer this puzzling query, let’s briefly
recall what’s been learned so far about the actual practice of
science in today’s world.

Our deliberations up to now allow us to summarize compactly
the practice as opposed to the philosophy of science in the follow-
ing two principles:

A. There is an ideology of science consisting of a cognitive struc-
ture (facts -* hypothesis experiment -* laws -» theory),
together with the processes of verification and peer review.

FAITH, HOPE, AND ASPERITY

B. Science is a social activity , with the standards for what con-
stitutes good science determined by the norms of a particular
community.

With these facts of modern scientific life in mind, let me now
offer a short checklist of “sights and sounds” (and smells) for
detecting pseudoscience. If you’re reading a paper and catch the
whiff of even one of the items on this list, be assured that the
author is dealing in pseudoscience, at least by the standards pre-
vailing in today’s world of science. For the following list I am
indebted to the outstanding work Science and Unreason by Mi-
chael and Daisie Radner, to which I direct the reader’s attention
for a far more extensive account of the whole culture of pseudo-
science and pseudoscientists.

HALLMARKS OF PSEUDOSCIENCE

• Anachronistic thinking: Cranks and pseudoscientists often re-
vert to outmoded theories that were discarded by the scientific
community years, or even centuries, ago as being inadequate.
This is in contrast to the usual notion of crackpot theories as
being novel, original, offbeat, daring, and imaginative. Good
examples of this kind of crankishness are the creationists, who
link their objections to evolution to catastrophism, claiming
that geological evidence supports the catastrophic rather than
uniformitarian view of the kind of geological activity they as-
sociate with evolution. The argument is anachronistic insofar
as it presents the uniformitarianism-catastrophism dichotomy
as if it were still a live debate.

• Seeking mysteries: Scientists do not set out in their work to
look for anomalies. Max Planck wasn’t looking for trouble
when he carried out his radiation emission experiments and
Michelson and Morley certainly were not expecting problems
when they devised their experiment to test for the luminifer-
ous ether. Furthermore, scientists do not reject one theory in
favor of another solely because the new theory explains the
anomalous event. On the other hand, there’s an entire school of
pseudoscience devoted to enigmas and mysteries, be they the
Bermuda Triangle, UFOs, yetis, spontaneous combustion, or
other even more offbeat phenomena. The basic principle under-
lying such searches seems to be that “there are more things in
heaven and earth than are dreamt of in your philosophy,” cou-

PARADIGMS LOST

pled with the methodological principle that anything that can
be seen as a mystery ought to be seen as one.

• Appeals to myths: Cranks often use the following pattern of
reasoning: Start with a myth from ancient times and take it as
an account of actual occurrences; devise a hypothesis that ex-
plains the events by postulating conditions that obtained at
that time but that no longer hold; consider the myth as provid-
ing evidence for support of the hypothesis; argue that the hy-
pothesis is confirmed by the myth as well as by geological,
paleontological, or archaeological evidence. This is a pattern of
circular reasoning that is absent from the blackboards and
laboratories of science.

• A Casual approach to evidence: Pseudoscientists often have the
attitude that sheer quantity of evidence makes up for any de-
ficiency in the quality of the individual pieces. Further,
pseudoscientists are loath ever to weed out their evidence, and
even when an experiment or study has been shown to be ques-
tionable, it is never dropped from the list of confirming evi-
dence.

• Irrefutable hypotheses: Given any hypothesis, we can always
ask what it would take to produce evidence against it. If noth-
ing conceivable could speak against the hypothesis, then it has
no claim to be labeled scientific. Pseudoscience is riddled with
hypotheses of this sort. The prime example of such a hypothe-
sis is creationism; it’s just plain not possible to falsify the
creationist model of the world, as we’ll see in the next chapter.

• Spurious similarities: Cranks often argue that the principles
that underlie their theories are already part of legitimate sci-
ence, and see themselves not so much as revolutionaries but
more as the poor cousins of science. For example, the study of
biorhythms tries to piggyback upon legitimate studies carried
out on circadian rhythms and other chemical and electrical os-
cillators known to be present in the human body. The basic
pseudoscience claim in this area is that there is a similarity
between the views of the biorhythm theorists and those of the
biological researchers, and therefore biorhythms are consistent
with current biological thought.

• Explanation by scenario: It’s commonplace in science to offer
scenarios for explanation of certain phenomena, such as the
origin of life or the extinction of the dinosaurs, when we don’t
have a enough data to reconstruct the exact circumstances of

FAITH, HOPE, AND ASPERITY

the process. However, in science such scenarios must be con-
sistent with known laws and principles, at least implicitly.
Pseudoscience engages in explanation by scenario alone, i.e.,
by mere scenario without proper backing from known laws
and theories. A prime offender in this regard is the work of
Velikovsky, who states that Venus’s near collision with the
Earth caused the Earth to flip over and reverse its magnetic
poles. Velikovsky offers no mechanism by which this cosmic
event could have taken place, and the basic principle of deduc-
ing consequences from general principles is totally ignored in
his “explanation” of such phenomena.

• Research by literary interpretation: Pseudoscientists frequently
reveal themselves by their handling of the scientific literature.
They regard any statement by any scientist as being open to
interpretation, just as in literature and the arts, and such
statements can then be used against other scientists. They
focus upon the words, not on the underlying facts and reasons
for the statements that appear in the scientific literature. In
this regard, the pseudoscientists act like lawyers gathering
precedents and using these as arguments, rather than attend-
ing to what has actually been communicated.

• Refusal to revise: Cranks and crackpots pride themselves on
never having been shown to be wrong. It’s for this reason that
the experienced scientific hand never, under any circum-
stances, enters into dialogue with a pseudoscientist. But im-
munity to criticism is no proof of success in science, for there
are many ways to fend off attacks: Write only vacuous mate-
rial replete with tautologies; make sure your statements are so
vague that criticism can never get a foothold; simply refuse to
acknowledge whatever criticism you do receive. A variant of
this last ploy is a favorite technique of pseudoscientists: They
always reply to criticism, but never revise their position in
light of it. They see scientific debate not as a mechanism for
scientific progress but as an exercise in rhetorical combat.
Again the creationists serve as sterling testimony to the power
of this principle.

The major defense of pseudoscience is summed up in the state-
ment “Anything is possible,” the pseudoscientific version of

Feyerabend’s philosophical theme song “Anything Goes.” Ear-
lier we considered the question of competition between models

PARADIGMS LOST

and theories and drew up a few ground rules by which the com-
petition is generally carried out in legitimate scientific circles.
Let’s look at how pseudoscientists, with their “Anything is pos-
sible” shield, enter into such competition.

In the competition among theories, the pseudoscientist makes
the following claim: “Our theories ought to be allowed into the
competition because they may become available alternatives in
the future. Scientists have been known to change their minds on
the matter of what is and is not impossible, and they are likely
to do so again. So who’s to say what tomorrow’s available alter-
natives may be?” In other words, anything is possible! The fact
that a theory may become an available alternative in the future
does not constitute a reason for entering it in the competition
today. Every competitor now must be an available alternative
now. The pseudoscientist suggests that we may as well throw
away the current scientific framework since it will eventually
have to be replaced anyhow.

By referring to a future but as-yet-unknown state of science,
the cranks are in effect refusing to participate in the competi-
tion. This would be all right if they didn’t at the same time in-
sist on entering the race. It’s as if one entered the Monaco
Grand Prix with a jet-propelled car and insisted on being al-
lowed to compete because, after all, someday the rules may be
changed to make it a jet car race!

The pseudoscientists also worm their way into the competition
by putting the burden of proof on the other side. They declare
that it’s up to the scientific community to prove their theory
wrong, and that the theory must be taken seriously if the com-
munity cannot do so. The obvious logical flaw is the assumption
that failing to prove a theory impossible is the same thing as
proving it possible. While the principle of innocent till proven
guilty may be used in Anglo-Saxon courts of law, scientific de-
bate is not such a court. The reason why pseudoscientists think
they can put the burden of proof on the scientists can be traced
to a mistaken notion of what constitutes a legitimate entry in
the debate. They think that the scientific method places a duty
on the scientific community to consider all proposed ideas that
are not logically self -contradictory. In their view, to ignore any
idea is to be prejudiced.

Finally, we note that the pseudoscientists often act as if the
arguments supporting their theory were peripheral to the the-

FAITH, HOPE, AND ASPERITY

ory. Science is defined in terms of how and why we know some-
thing, not what we know. Thus, the pseudoscientists fail to see
that what makes a theory a serious contender is not just the
theory, but the theory plus the arguments that support it.
Cranks think that somehow the theory stands on its own, and
that the only measure of its merit for entering the competition is
its degree of daring and novelty. Hence, they think the scientific
community has only two choices: admit their theory into the
competition or else prove it to be wrong. However, when it comes
to defending a theory or model in scientific debate, without high-
quality supporting evidence and a solid conceptual scheme,
there’s just no time, room, or patience for the “Anything is pos-
sible” antics of pseudoscience.

As a postscript on the pseudoscientists, it’s of interest to ask
why the ideas of many pseudoscientists like Velikovsky are so
popular. While it is true that Velikovsky ’s concepts are a little
simpler than those used by modern astronomers and paleontolo-
gists, his real advantage is that they are so much easier to visu-
alize mentally and come to terms with. In short, they appeal to
what John Q. Public would call common sense. Unfortunately,
neither the world nor science is as simple as naive common sense
would have us believe. For example, what kind of peasant cun-
ning would suggest that energy levels in atoms can come only in
discrete packages? Common sense would say that if you can
walk up stairs one step at a time, then you can also stroll up a
ramp to get to the same place. But modern physics says no:
Change of energy levels can occur only in discrete steps. The
more developed a scientific specialty becomes, the less reliable
common sense is as a guide. In fact, there are aspects of science
that are just plain contrary to common sense, like the staircase
example just noted. The point to keep in mind is that most be-
liefs being promoted as alternatives to science are deliberately
calculated to fit smoothly into what common sense suggests is
the way things should be, as well as the way to solve all our
problems. Within these comforting world views, we have no
problems of our own — everything that happens to us does so be-
cause of bad aspects of Jupiter, the work of the devil, or the will
of superior beings from Andromeda. At root, these beliefs are a
measure of the degree of disappointment with which the general
public greets the revelations of modern science. The average man
wants complete, easy-to-understand, clear-cut answers, when all

PARADIGMS LOST

that science has to offer is arcane, difficult-to-follow ifs, ands,
buts, or maybes.

Belief systems outside science come in many forms, some of
them covered by the general umbrella of pseudoscience. By far
the most interesting and important alternative to a scientific or-
dering of the world is that provided by the principles and tenets
of organized religion. From the beginnings of Western science
in the Middle Ages, there has been a sort of (not always unde-
clared) guerrilla war waged between the Church and the scien-
tific community on the matter of which is the keeper of true
knowledge about the nature of the cosmos. In the next section we
will examine this conflict as our final statement about the alter-
native realities that we use to shape and interpret our daily
lives.

THE PULPIT AND THE LAB

A few years ago Daysi Fernandez, a mother of three living on
welfare in New York City, bought a lottery ticket that came up
a winner, returning almost $3 million, a tidy profit on a $4 in-
vestment. Little did Mrs. Fernandez realize that in her good for-
tune she would become embroiled in a classic case pitting the
claims of science against those of religion. As the story goes,
Mrs. Fernandez had asked a young friend, John Pando, to pur-
chase lottery tickets for her. Pando, a staunch believer in the
power of prayer, thought that the chances of success for one of
the tickets would be greatly enhanced if he asked for the divine
intervention of Saint Eleggua. Apparently Mrs. Fernandez was
sympathetic to his beliefs, for he claimed that she had promised
to give him half the proceeds if any of the tickets struck gold. If
you’ve already guessed the punch line of this story, you’re just a
bit ahead of me.

One of Mrs. Fernandez’s tickets was drawn to the tune of
$2,877,203.30, but she refused to fork over the promised half of
the pie to Pando. In the tried and true American fashion for
dealing with such slights, Pando ’s immediate response was to file
a lawsuit against her, in an attempt also to gain entry to the
Millionaires’ Club. Mrs. Fernandez argued that the agreement
was illegal and/or unenforceable on a number of grounds, in-
cluding the fact that John Pando was a minor under the age of

FAITH, HOPE, AND ASPERITY

eighteen. After hearing the competing arguments, Judge Ed-
ward Greenfield of the New York County Supreme Court ruled
on the matter.

The judge found in favor of Pando on most of the points, in-
cluding the matter of age, but came up with a final verdict in
favor of Mrs. Fernandez on the grounds that it was impossible
in a court of law to prove that “faith and prayers brought about
a miracle and caused the defendant to win.” In other words,
Pando hadn’t proved that Saint Eleggua had rigged the lottery
to point the finger of fate at Mrs. Fernandez. As far as it goes,
this seems a defensible statement. But what is open to serious
debate is the reasons given by the judge for denying Pando a
share of the fortune.

Judge Greenfield in effect assumed a priori that religious be-
liefs are not amenable to scientific testing. As part of his deci-
sion, the judge also stated that rainmaking by cloud seeding
would qualify for payment, but that the production of rain by
dances, chants, and the other tricks of the medicine man’s trade
would not. Thus, the Fernandez case opens up for further in-
spection the age-old question of where a belief system stops and
science begins.

In the Reality Game, religion has always been science’s tough-
est opponent, perhaps because there are so many surface
similarities between the actual practice of science and the prac-
tice of most major religions. Let’s take mathematics as an exam-
ple. Here we have a field that emphasizes detachment from
worldly objects, a secret language comprehensible only to the
initiated, a lengthy period of preparation for the “priesthood,”
holy missions (famous unsolved problems) to which members of
the faith devote their entire lives, a rigid and somewhat arbi-
trary code to which all practitioners swear allegiance, and so
on. These features are present in most of the sciences as well,
and bear a striking similarity to the surface characteristics
of many religions. Both scientific and religious models of the
world direct attention to particular patterns in events and re-
structure how one sees the world. But at a deeper level there
are substantial differences between the religious view and that
of science.

Let’s consider some of the major areas in which science and
religion differ:

PARADIGMS LOST

• Language: The language of science is primarily directed to-
ward prediction, explanation, and control; religion, on the
other hand, is an expression of commitment, ethical dedica-
tion, and existential life orientation. So even though there are
superficial similarities at the syntactic level, the semantic con-
tent of scientific and religious languages are poles apart.

• Reality: In religion, beliefs concerning the nature of reality
are presupposed. This is just the opposite of the realist view of
science, which is directed toward discovering reality. Thus re-
ligion must give up any claims to truth, at least with respect
to any facts external to one’s own commitment. In this regard,
the reality content of most religious beliefs is much the same
as in the myths considered earlier. Fundamentally, what we
have in science is a basic belief that the universe is under-
standable using rational arguments, experimental observa-
tions, even divine inspirations, but no acts of blind faith. This
is a viewpoint that is not necessarily shared by many religions.

• Models: While both scientific and religious models are analogi-
cal, and used as organizing images for interpreting life experi-
ences, religious models also serve to express and evoke
distinctive attitudes, as well as to encourage allegiance to a
way of life and adherence to policies of action. The imagery of
religious models elicits self -commitment and a measure of ethi-
cal dedication. These are features completely anathematic to
the role of models in science. In religion the motto is “Live by
these rules, think our way, and you’ll see that it works.” The
contrast with the traditional ideology of science is clear.

• Paradigms: In the discussion of paradigms, we saw that scien-
tific paradigms were subject to a variety of constraints like
simplicity, falsification, the influence of theory on observation,
and so forth. All of these features are absent in the selection
of a religious paradigm.

• Methods: In science there is a set of procedures to get at the
scheme of things: observation, hypothesis, experiment; in reli-
gion there is a method, too — divine enlightenment. However,
the religious method is not repeatable, nor is it necessarily
available to every interested investigator.

Table 1.3 displays a comparative chart of the different ways of
science and religion. How are we to divine what this table is try-
ing to convey about the respective abilities of science and reli-
gion to tell us anything useful about ourselves and the universe

FAITH, HOPE, AND ASPERITY

ISSUE

RELIGION

SCIENCE

subject matter

God and humankind

phenomena of Nature

information source

revealed word, holy
books

observations,

experiments

objective of study

purpose and plan

mechanisms

language

everyday speech

mathematics

method

literary

interpretations

measurement and
analysis

results

moral imperatives

explanations

validation

personal experience

replication, testing

limitations

mechanisms

unexplained

no goals or values

community

church

scientific establishment

TABLE 1.3 Religion compared with science

we inhabit? It seems that there are at least three possible an-
swers to this classic conundrum:

1. Two realms: Science and religion have different spheres of ju-
risdiction.

2. Concordance: Religious and scientific explanations of Nature
can be brought together on the same plane.

3. Partial views: Science and religion each illuminate the same
reality (whatever that might be), but from different perspec-
tives.

To my mind, only the last possibility makes any sense whatso-
ever. The first leads to the all too depressing territorial disputes
of the kind that so much blood has been shed over through the
years, while the second is self-defeating since scientific views are
always changing. As a result, a theology that attaches itself to
one scientific family today will surely be an orphan tomorrow.

With the above considerations on religion under our belt, we
see that both pseudoscience and religion provide alternate real-
ity-structuring procedures radically different in character from
those employed in science. It’s of interest to ponder why there is
such a diverse mixture of nonscientific knowledge, especially in
view of the claims of virtually every sect that its own brand of
medicine is uniformly most powerful.

My view on this matter is quite simply that neither science nor

PARADIGMS LOST

religion nor pseudoscience gives a product that is satisfactory to
all customers; the wares are just not attractive enough. In some
cases the beliefs are not useful in the way that people want to
employ them. For example, many people have a deep-seated psy-
chological need for security and turn to conventional religion for
myths of all-powerful and beneficent Beings who will attend to
this need for protection. Science, with its mysterious and poten-
tially threatening pronouncements about black holes, the “heat
death” of the universe, evolution from lower beings, nuclear
holocaust, and so on, offers anything but comfort to such primal
needs and, as a result, loses customers to the competition. Basi-
cally, beliefs thrive because they are useful, and the plain fact is
that there is more than one kind of usefulness.

To the practicing scientist, the foregoing observations come as
a sobering if not threatening conclusion, since they seem to put
in jeopardy the conventional wisdom that the road to truth lies
in the “objective” tools of science, not the subjective, romantic
notions of believers and crusaders. But if we accept Feyera-
bend’s arguments of alternative and equally valid belief systems,
then we are inexorably led full circle back to the position that
there are many alternative realities, not just within science itself
but outside as well, and the particular brand of reality we select
is dictated as much by our psychological needs of the moment as
by any sort of rational choice. In the final analysis, there are no
complete answers but only more questions, with science provid-
ing procedures for addressing certain important and interesting
classes of such questions.

INTO THE COURTROOM OF BELIEFS

The British philosopher John Locke appears to have been the
first to use the word “science” in anything like its modern mean-
ing when he equated “scientifical” with certainty and demon-
stration of knowledge about the physical world. In the chapters
that follow, we will be out to question the degree to which science
delivers on these lofty aims. Our dual philosophical themes cen-
ter about the eternal puzzles: What is real, and what is our rela-
tionship as human beings to this reality? In the course of
attempting to shed light on these Bobbsey Twins of philosophi-
cal speculation, I have chosen the vehicle of a courtroom meta-

FAITH, HOPE, AND ASPERITY

phor within which the competing scientific (and sometimes pseu-
doscientific and/or religious) parties can plead their case. My
reasons for this setting are best summed up in the remark by
Henry Bauer that “where eminent men disagree violently, and
both sides present their cases as proven, we can be rather sure
that certainty is not in fact available, and that the matter is not
technical but rather trans-scientific. It is a dispute over proba-
bilities, values, desirability, not over facts.” The only factor that
characterizes science as a whole is that, in the long run, untruths
are weeded out and what remains becomes more reliable. Thus,
just as in economics where Adam Smith’s Invisible Hand guides
the flow of events into progressive channels, in science we have
the Invisible Boot, which acts to kick out those ideas, theories,
and beliefs that don’t prove useful to enough people over a suf-
ficiently long period of time.

I leave it to the reader to be the final judge of whether or not
“scientism” (I promise that this will be my last -ism) establishes
a case for its underlying thesis that “science = truth.” But suc-
ceed or fail, I hope that as we go through the various case stud-
ies in scientific conflict that follow, the reader will not only get
some basic grasp of the ideas themselves, but even more impor-
tant will discover that these ideas are genuinely worth an at-
tempt to understand them. Only by acquiring a deeper feeling
for the processes as well as the results of science will it be possi-
ble to assess its merits effectively as a reality-generation activ-
ity. So now that the anthems have been sung, the pledges of
allegiance given, and the witnesses called, the court is ready to
hear the first case in the continuing litigation between science
and Nature. Let the opening arguments proceed!

A W A R M

LITTLE POND

CLAIM:

LIFE AROSE OUT OF NATURAL PHYSICAL
PROCESSES TAKING PLACE HERE ON EARTH

OUT OF THE FIRE AND INTO THE SOUP

By most standards of comparison, 1953 was an eminently for-
gettable year, with only a fifth straight World Series triumph
for the Yankees, the death of Stalin, and Secretary of Defense
Charles Wilson’s immortal remark that “I thought what was
good for our country was good for General Motors, and vice
versa” brightening up what was otherwise a pretty dull trip
around the Sun. But not so in the world of biology; in fact, for
biologists 1953 was a vintage year the like of which had not been
seen since the publication of Darwin’s classic in 1859. In this

A WARM LITTLE POND

single twelve-month span, not only did Watson and Crick un-
ravel the double-helix geometry of DNA and Frederick Sanger
work out the chemical structure of proteins, but also the modern
era of scientific investigation of the origin of life on Earth was
ushered in with Stanley Miller’s experiment showing that the
chemical building blocks of life could be formed by natural
physical processes taking place in the primordial environment.
While the work of Watson, Crick, and Sanger is crucial for un-
derstanding how living forms function, it was Miller’s experi-
ment that set the stage for what has become the dominant
scientific paradigm for how life as we know it today got its start
here on Earth. To trace that thread, we must begin in 1923 in
Moscow with the unheralded publication of a booklet asserting
that there is no fundamental difference between living and non-
living matter.

Having just escaped the yoke of the czars and not yet stuck
their necks into the noose of Stalinism, Russians found the
Roaring Twenties to be an excellent decade for challenging es-
tablished orthodoxies. So it seems appropriate that during this
time a thirty-year-old biologist, Alexander I. Oparin, should
present the first real scientific case against biblical creationism,
arguing that life could have arisen by natural physical means
here on Earth. The essence of Oparin’s argument, later ex-
panded upon in his 1936 book Origin of Life, was that geological
evidence suggests the atmosphere of the early Earth was filled
with gases like methane, ammonia, hydrogen, and water vapor —
but no oxygen (i.e., it was what chemists call a reducing mixture ).
By pumping energy from lightning, ultraviolet radiation, vol-
canic heat, and natural radioactivity into such a blend of gases,
Oparin reasoned, the chemical components composing all living
things could be formed in the sea, ultimately accumulating to a
density at which they could link up to form the first primitive
living entities. A few years later, the British biologist J.B.S.
Haldane independently proposed the same general idea, color-
fully expressing the character of such a primordial sea as a kind
of “hot dilute soup,” leading to the modern labeling of this Opa-
rin-Haldane Hypothesis as the Primordial Soup Theory.

It’s of perhaps more than just passing curiosity to note that
both Oparin and Haldane were professed Marxists in a revolu-
tionary era when it was fashionable to try to solve all sorts of
problems here and now by dialectical and materialistic means.

PARADIGMS LOST

Oparin, who died in 1980, has been described as the kind of man
who at dinner had a bottle of cognac on one side and a bottle of
vodka on the other, both of which were empty by the end of the
meal. While the two-fisted drinking stories may or may not be
apocryphal, there is no doubt about Oparin’s unfortunate politi-
cal alliance with the crankish, but powerful, geneticist T. D.
Lysenko. During Lysenko’s tyrannical period of grace, Oparin
reigned over the Biology Division of the Soviet Academy of
Sciences, using his political clout to set Soviet biology back at
least twenty years. Finally, upon the death of Stalin and the
consequent decline of Lysenko, both Oparin and Lysenko were
removed from their administrative posts at the academy and re-
turned to the laboratory bench— Oparin to direct the Bakh In-
stitute of Biochemistry (whose main activity involved the study
of fermentation for making beer), Lysenko to continue his defi-
nite, but almost totally meaningless, Lamarckian experiments in
changing winter wheat to spring wheat. An indication of the
kind of political cunning that enabled Oparin to survive in such
a shifting political environment is displayed in his remark to the
journalist Harold Hayes, who, in a visit with Oparin shortly
before his death, asked about his views on the treatment of the
physicist and human rights advocate Andrei Sakharov. Oparin
replied, “Of course, there are many Sakharovs in Moscow!” In-
cidentally, Haldane, who for many years served as the UK edi-
tor of the The Daily Worker, the newspaper of the Communist
party, ultimately lost faith in Marxism for exactly the same rea-
son that Oparin was catapulted to power — Lysenkoism. As we’ll
see later, even though they were the co-originators of the Pri-
mordial Soup Theory, the two also stood on opposite sides of the
fence on the details of exactly how life actually came crawling
out of the broth. But we’re getting ahead of our story. Let’s now
go back to Stanley Miller and the Chemistry Department at the
University of Chicago in the early 1950s.

At the time, Miller was a young graduate student in the de-
partment shopping around for a Ph.D. thesis topic. Initially he
decided to do a theoretical project with Edward Teller, some-
thing involving the manner in which chemical elements could be
synthesized in stars. However, Teller’s departure shortly there-
after to set up what is now the Lawrence Livermore National
Laboratory put an end to Miller’s plan, sending him scurrying
about for an alternate topic, as well as a new adviser. At this

A WARM LITTLE POND

point fate intervened in the form of another faculty member,
Harold Urey, who had earlier given a departmental seminar that
Miller had attended and listened to with great interest. Urey,
who won the Nobel Prize for chemistry in 1934 for the discovery
of deuterium (heavy water), had turned his attention to prob-
lems of the origin of the solar system, claiming that the atmo-
sphere of the early Earth would have been highly reducing, i.e.,
lacking any free oxygen. As a consequence, Urey argued that
such an atmosphere should be a good place to synthesize organic
compounds that could then form the necessary raw materials out
of which to assemble the first living organisms. Urey concluded
by suggesting that someone should do an experiment to test the
feasibility of the idea. Later, Miller reports, he pointed out to
Urey that Oparin had made the same suggestion in his book,
although Urey’s discussion was far more thorough and con-
vincing.

Following Teller’s departure for California, Miller told Urey
that he wanted to do the experiment on organic synthesis in a
reducing atmosphere for his thesis topic. Urey was initially
against the idea, primarily because he saw it as a speculative
project that could chew up lots of time and energy with no visi-
ble output — just the sort of project that a graduate student in-
tent on getting a degree should steer clear of. However, Miller
persisted and finally Urey relented, allowing Miller to begin the
experiment in the autumn of 1952. The rest is history. The es-
sence of the experiment is depicted in Figure 2.1.

To simulate the primordial atmosphere, Miller used a combi-
nation of methane (CH4), ammonia (NH,), water vapor (H20),
and hydrogen (H2). Energy input to the mixture was supplied
by a spark discharge simulating lightning, with the entire mix-
ture circulated through a cooling tube causing the gaseous com-
bination to condense in an imitation of rainfall. Heat was
supplied to the liquid in order to simulate evaporation in the
ocean. After some initial playing about with the parameters of
the apparatus (rate of heating, amount of the various gases, se-
quence of heat, spark, and condenser), Miller let the apparatus
run for a week, after which time the mixture was found to con-
tain significant amounts of the amino acids glycine and alanine,
two of the basic building blocks of protein, hence essential ele-
ments in all life. It’s interesting that, while admitting his plea-
sure at the outcome, Urey confessed his surprise at finding such

PARADIGMS LOST

a significant amount of these essential compounds, remarking
that his expectations for the experiment were to find “Beil-
stein,” i.e., a little bit of everything — a reference to the classic
set of over a hundred volumes listing every organic compound
that has ever been synthesized.

Publication of the experiment in Science in May 1953 gener-
ated considerable attention in the popular press, even resulting

A WARM LITTLE POND

in a Gallup poll in which 22 percent of the respondents did not
exclude the possibility of creating life in a test tube. An interest-
ing scientific postscript to Miller’s work is the fact that an ear-
lier experiment aimed at the same goal of creating some of the
building blocks of life had been carried out by Melvin Calvin in
Berkeley at the very time that Miller was completing his under-
graduate work there. Calvin had used a quite different atmo-
sphere consisting of water vapor and carbon dioxide, with an
energy source of alpha radiation obtained from the Berkeley cy-
clotron. Since Calvin’s atmosphere was oxidizing rather than re-
ducing, he obtained no organic compounds. A puzzling aspect of
this experiment was Calvin’s use of the oxidizing atmosphere,
since he explicitly referred to Oparin’s reducing atmosphere in
his paper yet didn’t use it in the experiment. In fact, this obser-
vation led to a critique published by Urey that ultimately moti-
vated him to suggest the experiment performed by Miller. A
final anomaly surrounding this whole business is that Miller
made no mention of Calvin’s work in his historical account of
the events leading up to his own experiment. Yet it seems un-
likely that he could have been unaware of it, since he had been a
student in the very same department at precisely the time when
the work was under way. But such are the vagaries of fate and
information flow in the academic world.

In the three decades since Miller’s pioneering work, many sim-
ilar experiments have been carried out using a variety of gase-
ous mixtures and different energy sources, each experiment
leading to a slightly different collection of organic end products.
The current Head Soup Chef is Cyril Ponnamperuma, director
of the Laboratory of Chemical Evolution at the University of
Maryland, in whose office, fittingly enough, there is a large,
Andy Warhol-style picture of a Campbell’s soup can carrying
the label “Primordial Soup.” Ponnamperuma, who originally
studied religion in India and then moved into chemistry, has
managed to combine his two interests in the comment that “God
must be an organic chemist.” This pithy remark compactly sum-
marizes the core of today’s conventional scientific wisdom as to
how life on Earth originated: The basic building blocks of life
were synthesized from simple chemical elements that were in
ample supply on the primitive Earth. The compounds thus gen-
erated then managed to combine somehow, eventually forming
the first living organisms. At this point, Darwinian evolution

PARADIGMS LOST

and natural selection could begin working their magic to gener-
ate the myriad complex living forms we see today.

A crucial aspect of the credibility of this picture is the very
long time span of more than 4 billion years that evolution has
had to play with to create the myriad organisms we see on Earth
today. To convert this almost unimaginable magnitude into more
human terms, look at the spare change in your pocket. Now en-
tertain the happy vision of receiving $40 million for each and
every cent you find. That’s a ratio of 4 billion to 1. Let’s mea-
sure it on another scale. Suppose you try covering the distance
between New York and New Orleans with postcards — stacked on
edge! That’s also 4 billion for you, and each card represents one
year that Nature has had at her disposal to put modern life to-
gether. As an irrelevant aside, when 4 billion is put in these
terms we start to appreciate the true enormity of budgetary and
trade deficits measured in the hundreds, or even thousands, of
billions of dollars.

While the above skeletal outline serves to underpin most scien-
tific investigations of the origin of life, the fun really begins
when it comes time to spell out the details of the precise mech-
anisms Nature used to breathe the spark of life into a haphazard
mixture of simple chemicals. The controversies rage on over
these matters and we shall examine the competing arguments
later. But to make sense of the various claims, it’s first neces-
sary to understand the structure and operation of living forms
in greater detail, as only then will we be in a position to appreci-
ate the many gauntlets that must be run by any viable theory of
the origin of life.

A CRASH COURSE ON HOW LIFE LIVES

By more or less general consensus nowadays, an entity is consid-
ered to be “alive” if it has the capacity to carry out three basic
functional activities: metabolism, self-repair, and replication.
The latter two functions refer primarily to the entity’s ability to
manufacture good, but not necessarily perfect, copies of itself,
while the first involves the quite different ability to synthesize
from the surrounding environment the materials needed to en-
sure the entity’s survival. In all known life forms on Earth,
these two jobs are carried out within the cell by distinct chemical

A WARM LITTLE POND

compounds and processes. The metabolic functions are the prov-
ince of the proteins, while reproduction is handled by the nucleic
acids DNA and RNA (with a little help from the proteins). The
work of Sanger and others has shown that all proteins used in
modern life forms are formed as chains of amino acids and, fur-
thermore, of the many types of amino acids there are only
twenty that are used by living organisms. The work of Watson,
Crick, and many others demonstrated that the nucleic acids are
also formed as long sequences of chemical compounds termed nu-
cleotides. Each nucleotide is composed of one of five bases — gua-
nine (designated G), adenine (A), cytosine (C), thymine (T),
and uracil (U) — surrounded by some sugar and phosphate bonds
for structural integrity.

The way the cell goes about its main chemical business of
manufacturing proteins involves a lot of unfamiliar terminology
and a number of steps. So before setting forth on a far-too-accel-
erated tour of this territory, it will be helpful to have a familiar
analogy available to picture the process. Let’s think about the
way a modern automobile company like Ford or GM goes about
manufacturing a car.

First of all there is a master plan, or blueprint, describing the
entire automobile, as well as the processes and materials needed
for its manufacture. This plan is usually kept under lock and
key somewhere in corporate headquarters. In order to build var-
ious subsystems of the car like the motor, gearbox, or suspen-
sion, the relevant sections of the master plan are copied and
dispatched to the corresponding manufacturing plants. Let’s
consider the case of building your car’s motor.

When the working copy of the motor blueprint arrives at the
plant, specialized workers identify the various components
needed for such things as the engine block, pistons, and valves.
These “transfer” workers then go to the stock room to gather
the requisite items. The necessary items are next given to “as-
sembly” workers whose task it is to put them together into the
main components of the motor. As each main component, such as
the block, camshaft, valves, or rings, is put in place, the “trans-
fer agents” and the “assemblers” continue to read the working
plan and carry out its instructions until they finally come to the
instruction “stop: The motor is complete.” The finished motor is
then sent on its way and the process begins anew with the manu-
facture of another motor. As we will see in a moment, the meta-

PARADIGMS LOST

bolic machinery of the cell functions in a completely analogous
fashion, with its own version of master plans, working blue-
prints, transfer agents, and all the rest. Let’s see how it goes.

The cells of higher organisms are divided into two main com-
partments: the nucleus , which contains the cellular hereditary
“program,” DNA; and the cytoplasm, where the proteins are
manufactured. The real work of the cell is carried out by the
proteins, mainly the ribosomes, with the nucleic acids DNA and
RNA being a bit like the queen bee in a hive, fit only for repro-
duction but no real work. The DNA is divided into short sec-
tions, each of which represents either the chemical code for a
certain protein, or a control code that activates or inhibits cer-
tain chemical operations in the cell. Such sections of DNA are
called structural or regulatory genes, and they carry the informa-
tion needed to make the organism, as well as serving to pass that
information along to subsequent generations of cells. In simple
organisms like bacteria, there is only a single DNA strand, while
higher organisms contain a number of separate bundles of DNA
strands called chromosomes. The number of such strands varies
from species to species, being forty-six for humans, sixteen for
onions, and sixty for cattle. Some simple organisms like bacteria
and algae have no nucleus, with the genetic material mixed in
with the cytoplasmic material in a single compartment. Such
cells are termed prokaryotes (“cells without nuclei”). However,
virtually all multicelled organisms are composed of eukaryotic
cells, having a double-chambered structure with a separate
nucleus. The structure of such a cell is depicted in Figure
2.2., while Figure 2.3 displays the celebrated “double helix”
structure of DNA, showing the important base-pairing scheme
A <— ► T and C * G, together with the sugar and phosphate
bonds denoted by S and P. Note that the structure for RNA is
similar, except that there is only a single strand and the base
uracil replaces thymine. Now let’s briefly look at how such a cell
carries out its metabolic and reproductive activities.

Protein synthesis is initiated within the nucleus when a sin-
gle-strand “working copy” of part of the DNA, which can code
for one or more proteins, is made. This working copy is termed
messenger RNA (mRNA) and is formed by utilization of the sim-
ple base-pairing rules: Wherever the base A appears on the part
of the DNA strand that’s being copied, the mRNA strand will
have the base U, while if T appears on the DNA, the RNA

A WARM LITTLE POND

Nuclear Cell

membrane membrane

Nucleus

FIGURE 2.2 The structure of a modern eukaryotic cell

strand will show the base A. A similar pairing exists between the
bases C and G. Note here that the base TJ replaces the DNA base
T on RNA strands. Transcription is the technical term for this
process of copying part of the DNA onto a single RNA strand.
For eukaryotes, when the mRNA strand is complete it’s expelled
from the nucleus and is used in the cytoplasm as the program
for construction of the proteins called for by the genes it con-
tains.

The proteins are formed according to the following procedure.
Special combinations of proteins and RNA in the cytoplasm
called ribosomes move along the strand of mRNA, reading its ele-
ments (bases) in nonoverlapping groups of three. Each such
group is called a codon, and each codon is associated with either
one of the twenty amino acids of life, or with a “stop” signal,
according to the dictates of the genetic code. Since there are four
possible bases, and each codon consists of an ordered sequence of
three bases, there is a total of 4 X 4 X 4 = 64 possible codons.
The process of matching up codons with amino acids according
to the genetic code is termed translation, and the working-out of
this code constitutes one of the major triumphs of twentieth-
century biology. Since there are only twenty amino acids used in
forming proteins but sixty-four possible codons, we see that the
genetic code contains some measure of redundancy as all good

PARADIGMS LOST

t ' TT Thymine
UED Adenine
Guanine
E2 Cytosine
[s] Sugar

Deoxyribose
[p] Phosphate

FIGURE 2.3 The geometry of DNA

codes should. The precise correspondence between codons and
amino acids is shown in Figure 2.4. Note that the three codons
UAA, UAG, and UGA are “stop signs,” indicating to the ribo-
somes that they have come to the end of the program for that
protein.

Once the ribosome has read a particular codon, it must find
the corresponding amino acid in the cellular cytoplasm and join
it to the chain of amino acids already assembled from the earlier
codons. This process is carried out with the help of so-called
transfer RNA (tRNA). The tRNA is constructed in such a man-
ner that at one end it contains a “soeket” into which can be
plugged only one particular type of amino acid, while the oppo-
site end of the tRNA strand contains a sequence of three nucleo-
tide bases that form the anticodon to the amino acid at the other
end. (Technically speaking, these “sockets” are not at the ends
of the tRNA, but nearer the middle.) For example, if a strand of
tRNA holds the amino acid methionine (AUG) at one end, the
anticodon at the other end will be CAU, the mirror image of the
codon complementary to AUG. Note that it is the mirror-image
codon CAU and not the complementary codon UAC because the
two chains on the tRNA run in opposite directions. So if the
ribosome reads the codon AUG from the mRNA strand, it looks

A WARM LITTLE POND

phenylalanine

serine

tyrosine

cysteine

phenylalanine

serine

tyrosine

cysteine

leucine

serine

punctuation

leucine

serine

punctuation

tryptophan

leucine

proline

histidine

arginine

leucine

proline

histidine

arginine

leucine

proline

glutamine

arginine

leucine

proline

glutamine

arginine

isoleucine

threonine

asparagine

serine

isoleucine

threonine

asparagine

serine

isoleucine

threonine

lysine

arginine

methionine

threonine

lysine

arginine

valine

alanine

aspartic acid

glycine

valine

alanine

aspartic acid

glycine

valine

alanine

glutamic acid

glycine

valine

alanine

glutamic acid

glycine

FIGURE 2.4 The genetic code

for a tRNA molecule floating about in the cytoplasm having the
anticodon CATT, and when it finds it the amino acid methionine
is detached from the tRNA and added to the growing chain. At
this point, the tRNA has lost its amino acid and is dispatched
back into the cytoplasm where it looks for another unit of methi-
onine (in this case) to recharge itself. In this fashion, the ribo-
some moves along the mRNA strand assembling the protein

PARADIGMS LOST

chain one amino acid at a time, as if putting beads on a necklace.
When it comes to a “stop” codon it releases the protein chain
and begins work on another. The entire process is shown
schematically in Figure 2.5.

Since the foregoing cellular operation is so central to the ori-
gin-of-life debates, let’s try to fix the various steps and concepts
firmly by comparing them with corresponding elements in the
automobile-manufacturing analogy. The following chart shows
the match-ups:

cell

—

auto-manufacturing company

nucleus

corporate headquarters

cytoplasm

manufacturing plant (including stock room)

DNA

— 1 •

master blueprint for the car

mRNA

' •

working copy of part of the master blueprint

tRNA

— ♦

transfer workers and stockboys

ribosomes

assembly workers

structural gene

— *

plan for making a main component (e.g., motor)

regulatory gene

<—

plan for assembling the main components

amino acid

— *

individual part for a component

codon

«— •

blueprint ID number of an individual part

genetic code

4 *

rule for matching up parts with ID numbers

Of course, the foregoing analogy applies only to cellular metabo-
lism; reproduction comes extra. While no one has yet started de-
signing auto firms that literally reproduce themselves, the
reader should have no trouble seeing how to translate the cellu-
lar reproduction process into a corresponding program for a
self-reproducing car company.

The process of cellular reproduction is simplicity itself, being
almost self-evident from the picture of DNA given above. From
the base-pairing rules, it’s evident that if we had just one of the
two strands forming the DNA molecule, plus a good supply of
the various nucleotide bases, there would be no problem in recon-
structing the original double helix: just pair up the bases ac-
cording to the rule: A <— ♦ T and G ♦— > C. In real DNA this is
very close to the procedure actually followed, as the DNA is un-

A WARM LITTLE POND

PARADIGMS LOST

wound by enzymes (special-purpose proteins) while other en-
zymes act to link up the newly formed strands of DNA accord-
ing to the foregoing base-pairing rule.

While each of the above stations on the road of life is simple
enough on its own, there are quite a number of them, and keep-
ing them straight, together with the perhaps unfamiliar termi-
nology, is quite a task. As an aid, the box below provides an
oversimplified, but adequate for our purposes, summary of the
most important steps.

DNA || — >

mRNA

tRNA + mRNA flltog0”>es,

genetic code

Protein

The process of protein formation as described above is seen to
be a transfer of information in one direction: from the genetic
program encapsulated in the DNA to the proteins assembled in
the cytoplasm. In 1958 Francis Crick, codiscoverer of the dou-
ble-helix geometry of DNA, summarized this information flow in
what he termed the Central Dogma of Molecular Biology. This
“dogma” can be summarized by the diagram:

DNA t?”Cfiption RNA translation protein

The degree to which the arrows represent inviolable directions of
information flow has been a hotly debated point in molecular bi-
ology ever since Crick made his pronouncement. Instances of in-
formation passing from RNA back to DNA are known, but a
transfer from proteins back to either of the nucleic acids would
call for a major rethinking of the entire mechanisms of heredity
resulting in, among other things, a revival of the now-unfashion-
able idea of Lamarckian inheritance, i.e., the inheritance of ac-
quired characteristics.

Part of the tenacity with which biologists cling to the Central
Dogma is surely accounted for by Crick’s choice of the word
“dogma,” signifying a definite and authoritative doctrine. In ad-
dition, Crick’s reputation (and Nobel Prize) for his DNA work no
doubt also contributed its share to the dogma’s entrenchment in
the minds of biologists. Consequently, it’s one of science’s amusing
ironies that Crick himself later cheerfully confessed that he had
misunderstood the meaning of the word when giving the idea a
name, thinking that “dogma” referred only to “an hypothesis,

A WARM LITTLE POND

some arbitrary thing that was laid down for no particularly good
reason.” If naming the idea today, Crick claims, he would call it
the Central Hypothesis, clearly indicating that the notion is by no
means an established fact but only a provisional assumption or
working hypothesis. On this ambiguous note, let’s end our crash
course in the mechanisms of life and reexamine the Primordial
Soup Theory in light of what we have learned so far. But before
doing so, let’s pause just to summarize in the box below the bewil-
dering array of terminology introduced in this section as a point
of reference for the remainder of the chapter.

TERMS AND CONCEPTS

nucleic acid the genetic component of the cell, DNA or RNA.
Formed from the nucleotide bases, A, G, C, T, and U, plus
sugar and phosphate bonds

gene a short section of DNA that either codes for a single
protein or contains instructions regulating cellular chemi-
cal operations

mRNA a working copy of a segment of DNA used in the pro-
cess of gene translation into proteins

codon a triplet of nucleotide bases forming the source “lan-
guage” for the genetic code

tRNA a special-purpose molecule carrying an amino acid at
one end, its corresponding anticodon at the other end

ribosome a cellular “constructor,” which assembles proteins
by reading the codons from the mRNA and then linking
the relevant amino acids carried by the tRNA

cenetic code the rule by which codons are matched with one
of the twenty amino acids used by all living things

translation the process of protein assembly by ribosomes
reading the mRNA strand

transcription the process of producing the RNA strands
from DNA by the base-pairing rule

replication the process of producing a new DNA strand by
means of base-pairing rules

Central Dogma the claim that cellular information passes
only in the direction from the genes to the proteins

PARADIGMS LOST

POTHOLES ON THE ROAD TO LIFE

There are three evident facts about life that any decent theory
of its origins needs to confront:

Fact A: There is life on Earth.

Fact B: All life operates according to the same basic mech-
anisms.

Fact C: Life is very complicated.

To explain Fact A, a “soup theory” would have to show how the
conditions of the early Earth could give rise to living forms,
while explanation of Facts B and C would involve displaying a
plausible path for how primitive living organisms could ulti-
mately evolve the very complicated gene-protein symbiosis seen
in modern life forms, in addition to offering a convincing expla-
nation of why all living entities use the same small set of basic
chemical components and the same genetic code in carrying out
their life functions.

The most imposing hurdle that any origins theory must sur-
mount is the Gene-Protein Linkup Problem. As sketched above,
in order for the proteins to be “manufactured,” it’s first neces-
sary for the genetic material to be read and then decoded into
the appropriate amino acids according to the genetic code. On
the other hand, until the proteins are present there can be no
genetic material, since the process of replication is completely
dependent upon the activities of special proteins (replicases)
that facilitate the copying process. So we’re left with a real
“chicken and egg” situation, one that’s especially tricky for any
origins theory claiming that either the gene or the protein came
first, with the other following as a corollary. A number of inge-
nious arguments have been concocted to evade this vicious circle,
and we’ll examine them in detail later. But for now let’s just
briefly list some of the difficulties that origins theories must ad-
dress over and above the Gene-Protein Linkup Problem.

• Genetic code/protein structure: The laws of chemistry admit the
formation of hundreds, if not thousands, of distinct types of
amino acids; ditto for nucleotides. Why are all life forms
based on the use of only twenty amino acids and five types of

A WARM LITTLE POND

nucleotides? How did Nature settle on just these few chemical
types and why? Or equivalently, with so many types to choose
from, why isn’t it evolutionarily advantageous to make use of
the specialized properties of the other forms of amino acids to
make proteins, and the other kinds of nucleotides to construct
the genetic material? A related question relevant to the possi-
bility of extraterrestrial life is whether it’s necessary to use
separate molecular structures for the proteins and the nucleic
acids. On Earth, the proteins are good for action while the
chemical structure of the nucleotides is good for storing infor-
mation. But in some exobiological environment it might be
that the same chemical structures could be used for both pur-
poses.

• Chirality: Everything in Nature (except a vampire) has a mir-
ror image, and all the amino and nucleic acids come in both
left-handed and right-handed forms. While these two forms
are chemically identical in the sense of being formed from ex-
actly the same atomic constituents, the chemical actions of the
two forms are quite different as a result of their “twisting” in
opposite directions. In Miller-type experiments, approximately
equal quantities of both left- and right-handed molecules are
formed, and observations of the molecular composition of ga-
lactic clouds show a similar distribution between a given mole-
cule and its mirror image. Yet all life forms on Earth use
exclusively left-handed amino acids to form proteins and
right-handed nucleic acids to form the genetic material. As a
consequence of this puzzling fact, we could starve to death on
a planet where the steaks were made out of right-handed pro-
teins, since our body chemistry would be unable to break these
proteins down to extract their energy. A viable origins theory
would have to offer some coherent explanation of why living
forms settled exclusively on L (evo)-amino acids and d (extro)-
nucleotides, casting their mirror images aside.

• “Junk DNA”: It’s been observed that every strand of DNA
(other than in bacteria and viruses) contains long sections of
nucleotides that don’t code for any proteins. In other words,
reading along a strand of DNA would be similar to listening to
a typical transoceanic phone conversation in which only every
third or fourth word is actually comprehensible, the rest being
garbled or swallowed up by cosmic noise, crosstalk, and other
foibles of international communication circuits. These “junk”

PARADIGMS LOST

segments of DNA have to be edited out before the mRNA
strand leaves the nucleus to be used as the template for pro-
tein construction, and there are special editing enzymes in the
nucleus whose sole function is to perform just this task. While
it’s not strictly a question for origins theorists but for molecu-
lar evolutionists, it is still of interest to ask why Nature has
allowed this genetic “noise” to remain in the DNA. Or even
better, what’s this junk doing there in the first place? It
clearly serves no useful purpose in expressing the proteins
coded for in the DNA, since it’s removed during the process of
creating the messenger RNA strand used to construct the pro-
teins. Yet evolution has not seen fit to eliminate this noise
from the system, and it’s a puzzle for theorists to say why not.

Over twenty-five years ago, Howard Pattee noted that the best
way to experimentally test the claims of the primordial soup
theorists would be to create a completely unbiased prebiotic en-
vironment, then turn the system on and see what happens. In
Pattee’s words, “For all the inevitable inaccuracies in detail, a
sterile simulated seashore, with waves, tides, sand, rain, and in-
termittent sunlight, is a more accurate primitive Earth environ-
ment than the well-defined but oversimplified reactions studied
so far.” Recently N. Lahav and others have suggested that we
actually build a Whole Environment Evolution Synthesizer
(WEES), consisting of a combination of primary, secondary,
and tertiary environments open to various energy inputs. The
primary environments would consist of various gaseous combi-
nations thought to have composed the primitive Earth’s atmo-
sphere, while the secondary environments would be formed by
the primordial seas, lagoons, and ponds. The tertiary environ-
ments would then be composed of small fluctuations in these
more basic environments. In the WEES, the three phases (gas,
liquid, and solid) simulate the Earth’s biosphere using atmo-
sphere, sea, and land. The interfaces would include a tidal zone
and ponds, which are fluctuating environments. Material cycling
would take place within each environment, as well as between the
different environments. The main parameters used in controlling
the WEES would be the intensity, duration, and rhythm of en-
ergy inputs, the composition and pressure of the gas phase, and
the chemical and mineralogical composition of both the second-
ary and tertiary environments. A diagram of a WEES device is
shown in Figure 2.6.

A WARM LITTLE POND

ENERGY SOURCES

Gas

replenishment

Electrical UV-
Discharge Visible

Heat

Exchangers

ATMOSPHERE

Ponds

Land

Tidal

Sea

zone

FIGURE 2.6 A Whole Environment Evolution Synthesizer

Given the projected time span of many millions of years for
life to emerge, it’s wildly optimistic to expect Godzilla or even
primitive protozoa to come climbing up over the sides of a
WEES tank. Nevertheless, given the plethora of useful infor-
mation that’s emerged from Miller-type experiments, it’s not un-
reasonable to hope to learn substantially more details of how life
could have gotten its start by using the far more elaborate
WEES apparatus, including insight into some of the questions
raised above.

With the theoretical problems and Miller and WEES-style ex-
perimental apparatus as background, let’s now turn our atten-
tion to a more detailed consideration of the Prosecution’s case,
and look at the numerous variations on the Primordial Soup
Theory arguing in favor of an origin of life here on Earth by
natural chemical and physical means. Since their claims cur-
rently hold center stage, we begin with the arguments of those
asserting that the genes came first, everything else being a de-
tail.

PARADIGMS LOST

MONSTERS, HYPERCYCLES, AND
NAKED GENIES

In the mid-1960s, the biochemist Sol Spiegelman performed a
remarkable experiment. He placed a supply of the primitive Q/3
virus in a test tube together with a virtually inexhaustible sup-
ply of the replicase enzyme that the virus needs for replication
of its RNA. So that the virus would have no need to invade a cell
to complete its normal life cycle, Spiegelman also provided an
ample supply of free nucleotides in the tube. After mixing all
these ingredients together and arranging a continuous flow of
materials through the system, Spiegelman sat back to watch
what has come to be called evolution in a test tube. The original
RNA contained on the order of forty-five hundred nucleotides,
which coded for several proteins that the virus usually needed in
the wild to provide its protective coat, as well as to generate the
replicase enzyme required for its replication within a host cell.
But in Spiegelman’s setup none of these proteins were needed,
since the virus was insulated from external “predators” and was
being supplied with all the replicase necessary to reproduce at
whatever rate it wanted.

The outcome of the experiment was quite extraordinary. Ini-
tially, the naturally occurring Qy3 RNA copied itself more or less
faithfully. Rather quickly, however, mutations having the effect
of cutting the RNA strand in half occurred. Since it’s quicker
and easier to copy a short strand than a long one, such muta-
tions soon gained the upper hand in the Darwinian race for sur-
vival. As this process continued, shorter and shorter mutations
appeared, until after about seventy generations the system sta-
bilized at the shortest possible RNA strand capable of replica-
tion. It turned out that this strand contained about 220
nucleotides, and consisted of little more than the recognition site
for the replicase enzyme. This final form of the RNA was termed
the Spiegelman Monster, and offers an object lesson in the bad
things that can happen if life is too easy. This little monster was
able to reproduce itself at a staggering rate when confined to
the friendly environment of the test tube, but couldn’t possibly
hope to survive in the rough-and-tumble world of unprotected
reality.

A WARM LITTLE POND

Spiegelman’s experiment involved dumping a living Q/3 virus
into an artificially hospitable environment, consisting of a sup-
ply of free nucleotides and replicase enzymes. The Nobel-win-
ning German chemist Manfred Eigen took the process one step
further by omitting the “seed” virus. In Eigen’s experiment, a
supply of nucleotides and replicase enzymes was placed in a test
tube and left to its own devices. To everyone’s surprise, with no
seed virus to work with, the replicase enzyme proceeded to create
a short strand of RNA, showing that what’s important in the
experiment is the replicase enzyme, not the initial viral RNA.
The particular kind of RNA that emerged varied from experi-
ment to experiment, but all variations were close relatives of the
Spiegelman Monster and consisted of strands whose lengths
were about 120 nucleotides.

The experiments of Spiegelman and Eigen demonstrate the
minor gap of a hundred or so nucleotides that separates an RNA
molecule that grew out of nothing from one that began as part
of a living agent. This is a small difference indeed, and offers
ample testimony to how easy the process of replication really is.
Results of this sort supply the experimental muscle supporting
the claim of the so-called naked genies, theorists who believe that
the first living organisms were nothing more than short strands
of primitive RNA, consisting of a hundred or so nucleotides hav-
ing no purpose other than to perpetuate themselves. But there
are at least two major obstacles in the path of acceptance of
these claims, one involving the Gene-Protein Linkup Problem,
the other relating to the likelihood of such a replicator’s “self-
assembling” in the primordial ocean. Let’s examine these diffi-
culties in more detail in order to assess the plausibility of the
“genes first” arguments.

The essence of the naked genies claim is that the first living
things were random replicators that assembled themselves from
components floating around in the primordial soup. In particu-
lar, this means that there were no proteins, hence no replicase
enzymes. However, the sine qua non of both the Eigen and the
Spiegelman experiments was the presence of the replicase en-
zyme that facilitated the RNA replication. Thus, while these ex-
periments show that very small RNA strands are capable of
replication, they don’t begin to address the issue of how such
strands could ever arise without the help of the replicase. This
fact poses an enormous barrier for the naked genies to overcome,

PARADIGMS LOST

with the question currently being attacked on two different
fronts.

One line of attack is to try to make self -replica ting RNA
emerge without the assistance of a replicase. Using some artifi-
cially constructed, energy-rich nucleotide units, Leslie Orgel at
the Salk Institute in La Jolla, California, has managed to induce
RNA molecules to form a new chain that matches the existing
one, with the chain then forming into a double helix. Unfortu-
nately, the longest such chain has only about fifteen nucleotide
units, and the special units are of a type very unlikely to have
been present in the primeval seas. Furthermore, the replication
process stopped when the double-helix geometry formed; there-
after no additional RNA replication took place. For these rea-
sons, Orgel has been hesitant to make any claims about what he
terms his “models,” although others have asserted that these re-
sults show that it’s possible, in principle, for a naked gene to
replicate itself without benefit of a protein helper.

More recently, the work of Thomas Cech, Sydney Altman, and
others has shown that under plausible circumstances, it’s possi-
ble for RNA to act as an autocatalyst by snipping out a central
portion of itself and then resealing the cut ends. In addition,
they have shown that an RNA molecule can also cut up RNA
molecules different from itself, thereby acting as a true catalyst
(enzyme). Such self -catalytic RNA is also capable of joining sev-
eral short RNA molecules into a longer chain under conditions
that could possibly have been present on the early Earth. Fur-
ther experimentation along these lines has shown how it would
also be possible for RNA molecules to exhibit recombination, i.e.,
the ability to produce new combinations of genes, thereby pro-
viding the equivalent of sex — the infectious transmission of gen-
etic elements from one organism to another.

Walter Gilbert of Harvard, another Nobel-winning chemist,
has taken the above cluster of results involving self-catalytic
RNA and used them to construct a scenario for the origin of life
as we know it today, including a plausible explanation for the
earlier-noted junk DNA. Let’s look at the main steps.

THE GILBERT SCENARIO

A. RNA molecules perform the self-catalytic activity needed to
assemble themselves from the “soup.”

B. The RNA molecules evolve in self-replicating patterns, using

A WARM LITTLE POND

recombination and mutation to explore new functions and to
adapt to new niches.

C. The RNA molecules develop a range of enzymic activities.

D. RNA molecules begin to synthesize proteins, which are better
enzymes than their RNA counterparts; i.e., they perform the
same functions more efficiently.

E. Such protein enzymes are encoded by the RNA exon, the
part of modern DNA that is not edited out in the construc-
tion of the mRNA, i.e., the complement of the junk DNA.

F. Finally, DNA appears, giving a stable, error-correcting in-
formation store.

G. RNA is then shoved off center stage, having been replaced by
its creations, the proteins and DNA, which are able to per-
form its earlier double function more effectively.

The biggest question mark in Gilbert’s plan for the emergence of
life is Step A, since the experimental results on self-catalytic
RNA pertain only to the sophisticated present-day form of RNA,
and not to the presumably far more primitive forms of several
billion years ago. Thus, the problem still remains open as to the
degree to which self -catalysis of modern RNA sheds light on the
same possibility for more elementary forms.

Postulation of mechanisms for the random assembly of sim-
ple, primitive RNA chains also generates another set of diffi-
culties revolving about the amount of error tolerance that any
such “manufacturing operation” must accommodate. Examina-
tion of this issue leads to what we can term the Eigen scenario
for the origin of modern life. The basic idea consists of the fol-
lowing sequence of steps:

THE EIGEN SCENARIO

A. Start with a primordial soup consisting of randomly con-
structed small proteins, a sufficient quantity of lipids (fatty
acids) to be able to construct cellular membrane fragments,
and a variety of active, energy-rich nucleotide units suitable
for the construction of nucleic acids.

B. Assume that at least one replicating RNA molecule forms by
chance in the above soup. The assembly of such a molecule
could possibly have been assisted by the presence of proteins
that have also been randomly formed in the soup. Further-
more, this molecule is not a gene, as it codes for no protein; it

PARADIGMS LOST

is just a replicator. This molecule doesn’t have a unique nu-
cleotide sequence, but belongs to a family of closely related
individual molecules that Eigen calls a quasi-species.

C. In some manner RNA molecules then learn to exert control
over proteins, and a primitive genetic code develops. The dif-
ferent quasi-species specialize to take on different functions,
so that the entire population is capable of constructing a pro-
tein.

D. A series of complex and cooperative interactions now take
place between various nucleic acids and proteins. These in-
teractions have been termed hypercycles by Eigen, and have
been the subject of extensive mathematical and laboratory
analysis, which we’ll look into in a moment. The hypercycles
eventually gain control over their environment until they
reach levels straining the environmental carrying capacity.

E. At this point, to progress further it’s necessary for competi-
tion again to enter the picture. The lipids present in the ini-
tial soup are now utilized to construct compartments, each
compartment initially containing about the same mix of
quasi-species. However, as random mutations take place,
different types of hypercycles emerge, each contained in its
own membrane. These membranes compete with each other,
forming the prototypes of what later come to be modern
cells.

F. The processes of biological evolution now take over from
those governing the earlier chemical evolution, and ulti-
mately modern life forms evolve.

The Eigen scenario has the satisfying aspect that one general
principle, Darwinian evolution, is extended back to the time of
the first replicator. However, this scenario suffers from the same
defect as the Gilbert picture, namely, Step B: the appearance of
the first replicator. Eigen assumes this initial spark of life
emerges by nothing more than just a chance encounter of a set
of the right hundred or so nucleotides. Since this random assem-
bly problem lies at the heart of both the Gilbert and the Eigen
pathways to life, it’s worthwhile digressing for a moment to look
into its plausibility in somewhat more quantitative detail.

To illustrate the difficulty involved in randomly assembling
even a small RNA strand, suppose we have an organism that

A WARM LITTLE POND

reproduces asexually and is capable of producing ten offspring
before it dies. If the population is to be maintained without gen-
etic deterioration, at least one of the offspring must have the
same genetic information as its parent, while the other nine
could have mutations that would render them less fit to survive.
However, if not even one of the offspring is without mutation,
then the population will eventually decay and become extinct.
Suppose the RNA of this organism consists of 10,000 nucleotide
bases, and that these are replicated with an error rate of 1 per
1,000. Then the chance that all 10,000 bases are correctly repli-
cated is a paltry (999/1, 000 )10000, or about 1 in 22,000. So with
only ten offspring, there’s little chance that a population of such
organisms could long survive. As a rough rule of thumb, if a
population is to survive and has a chain of N nucleotide bases in
its genetic pattern, then it must have an error rate of less than 1
in N.

The above considerations lead us to ask how we humans with a
DNA strand many millions of bases long manage to replicate our
genetic patterns (genomes). The answer is that we have a “proof-
reading” stage, in which our replicase enzyme first puts in a
base with an error rate of about 1 in 10,000 and then checks it,
replacing it if it’s wrong. The second stage also has an error rate
of 1 in 10,000, so the overall error rate is a comfortable 1 in 100
million.

The dilemma for the Gilbert and Eigen scenarios is that their
primitive replicators have to make do without the replicase en-
zymes that provide the error-correcting step in replication, and
hence they have to put up with error rates in excess of 1 in 100.
As seen in the Spiegelman and Eigen experiments, this limits the
genome size to around a hundred bases. To improve upon this,
the primitive replicators would have to code for a replicase en-
zyme, as well as for a primitive protein-synthesizing machinery.
But that can’t be done with only a hundred bases. Thus, if you
can’t increase your genome size, you can’t code for an enzyme; if
you can’t code for an enzyme, you can’t increase your genome
size — Catch-22 for the naked genies.

The hypercycle concept is Eigen’s proposed solution to the
above dilemma. This notion relies upon the idea of dividing up
the genetic message to be copied into sections, and then imposing
natural selection on each section independently. The difficulty
with a straightforward use of this idea is that it’s unclear how

PARADIGMS LOST

to prevent one of the sections from outcompeting the others. If
all the sections are competing for the same bases, and if one re-
plicates faster than the others, then that fast-track replicator
will in time displace all the others and the resulting message will
consist only of the winning section of RNA. The hypercycle of-
fers a theoretical way out of this impasse.

Suppose the chain to be copied consists of the message A-B-
C-D, divided into the four sections A, B, C, and D. Imagine that
each of these sections represents a particular molecular popula-
tion, and that the populations are arranged in the hypercycle
shown below, with the rate of replication of each molecule in the
cycle depending upon the concentration of the molecule immedi-
ately preceding it in the sequence.

A -
T

D «-

Eigen and Peter Schuster have shown that if such relationships
exist, then the whole cycle is stable: No one molecule replaces all
the rest. Intuitively, the reason for this is that if the concentra-
tion of any molecule rises relative to the others, the net result is
to stimulate the others more than itself, the overall balance in
the cycle then being restored.

With the hypercycle structure, it’s possible to maintain and
replicate information selectively in an amount greater than
would be possible if the entire message A-B-C-D were copied as
a single unit. In their analysis of the mathematical properties of
such cycles, Eigen and his co-workers have shown that it’s possi-
ble for these cycles to evolve, with evolutionarily improved hy-
percycles more likely to emerge if the molecular quasi-species
are not able to move about too freely. This fact strongly sug-
gests the desirability of some sort of cellular membrane to con-
fine the components of the cycle.

To test the feasibility of the hypercycle scheme, IT. Niessert
conducted a series of computer experiments simulating the be-
havior of quasi-species and hypercycles according to Eigen’s
rules. She discovered that in addition to the error catastrophe dis-
cussed above, the molecular populations of a hypercycle are sub-
ject to at least three other types of catastrophes, colorfully
termed the selfish RNA, short circuit, and population collapse
catastrophes. The characteristic elements of each are:

A WARM LITTLE POND

• Selfish RNA: This situation occurs when a single RNA mole-
cule mutates to a form that replicates faster than its competi-
tors but, like some of my overly stimulated students, is having
so much fun replicating that it forgets its other role as a cata-
lyst.

• Short circuit: This catastrophe takes place when some RNA
molecule that’s supposed to be a link in the hypercycle chain
changes its role in such a way as to catalyze a later reaction in
the chain, thereby short-circuiting the cycle and contracting
the hypercycle into a simpler one.

• Population collapse: This brand of catastrophe happens when
statistical fluctuations result in the die-off of one of the molec-
ular species in the cycle, resulting in the collapse of the entire
chain of reactions.

Niessert discovered that the likelihood of the selfish RNA and
short-circuit catastrophes increases with the size of the molecu-
lar population, while, of course, the population collapse catastro-
phe is more likely with small species populations. Consequently,
the hypercycle model must sail a precarious path between the
Scylla of selfish RNA and short circuits and the Charybdis of
population collapse. There is only a narrow range of population
sizes for which the probability of all three catastrophes is low,
and even then the lifetime of a hypercycle can be shown to be
finite. These results tend to cast doubt upon any theory of the
origin of life that relies upon the cooperative organization of a
large population of molecules, especially if that theory provides
no insulating mechanisms to guard against the short-circuiting
of metabolic pathways. Despite their current preeminence as the
most popular flavor of the primordial soup, the naked genie ar-
guments all suffer from this glaring defect. So we now turn our
attention from the egg to the chicken, and examine the case for
the proteins-first theories.

THE CHICKEN’S STORY

Following a two-sentence summary of the Primordial Soup The-
ory in their 1974 book The Origins of Life on the Earth, Stanley
Miller and Leslie Orgel comment that “no one should be satisfied
with an explanation as general as this.” This remark aptly sum-

PARADIGMS LOST

marizes the view that any sensible skeptic would take of the
naked genie theories of the origin of life, motivating a considera-
tion of the other side of the coin: the possibility that the proteins
came first. On chemical grounds it’s not a bad bet to bank on the
proteins-first idea, since in Miller-type experiments it’s much
easier to form the amino acid building blocks of the proteins
than it is to generate the various sugars, phosphates, and nucleo-
tide bases needed for the nucleic acids, let alone to form a self-
replicating molecule like RNA. The problem, as we’ll soon see, is
that it’s very hard to construct any plausible scheme for the rep-
lication of proteins, other than through the nucleic acid inter-
mediaries that encode them. In this section, we’ll look at a couple
of the leading efforts devoted to ignoring this obstacle.

Historically, the idea that the first living forms were proteins
had the starring role at the very beginning of the scientific study
of life’s origins in the 1920s. This was the theory favored by
Alexander Oparin himself (although Haldane, the coorigina-
tor of the Soup Theory, was a genie). In a long series of ex-
periments, Oparin noted that if certain oily liquids are mixed
with water, it can happen that the oily liquid will form into
small droplets that then remain suspended in the water. These
small droplets are termed coacervates, and are reminiscent of
the tiny droplets of water forming a heavy mist or a pea-soup
fog, although they are of quite different composition. In one fa-
mous experiment, Oparin considered droplets formed from his-
tone (a protein) and gum arabic (a carbohydrate). When he
added an enzyme able to link sugars to form starches (the en-
zyme was, of course, obtained from some already living cell), the
enzyme accumulated in the coacervate droplets. He next added
glucose (a sugar) to the mixture, whereupon the sugar molecules
diffused into the droplets and combined to form starches that
remained within the droplet. As this process continued, the
droplets grew and eventually split, with each offspring droplet
also growing just as long as enzymes were continually added to
the mixture.

Superficially, Oparin’s coacervate droplets have a metabolism
as well as being able to grow and divide. But they’re able to do
so only because they’re being continually supplied with an en-
zyme from the outside, an enzyme synthesized by an already liv-
ing organism. Also, the droplets have no mechanism whatsoever
for replicating hereditary information; hence, they have no way
of evolving. Oparin apparently believed that life began by the

A WARM LITTLE POND

accumulation of more and more complicated molecular popula-
tions within the shells of these coacervate droplets. Evidently, he
felt that the external supply of the enzyme, which plays such an
integral role in his experiments, could be provided over the
course of geological time by natural processes occurring in the
primordial soup, and didn’t constitute a fatal stumbling block
for his basic vision of the origin of life via “oil droplets.” The
main steps in this vision are as follows: First the cellular mem-
branes form; then enzymes appear in order to organize the ran-
dom collection of molecular constituents in the broth into
metabolic pathways of various sorts; finally genes make their ap-
pearance. Since Oparin seemed to have only the haziest notion of
the role of genes, having carried out his work decades before
Watson and Crick, his theory of life basically says nothing
about these carriers of the hereditary message. We can summa-
rize Oparin’s view of life as:

OPARIN'S SCENARIO

Cells (Coacervates) -* Enzymes (Proteins) -*■ Genes

In 1963 at the unlikely location of Wakulla Springs, Florida,
the Second International Conference on the Origin of Life took
place, a gathering that provided the first and only opportunity
for Oparin and Haldane to meet face to face. The organizer
of that historic event was Sidney Fox, now at the University of
Miami, and a prime proponent of the proteins-first school of
thought on the origin of life. Fox has promoted the notion
of proteinoid microspheres, which were first discovered in his lab
in the 1950s, as the solution to the origins question. Since his
arguments have been favorably received by the media, as well as
winning honorable mention in several technical publications, it’s
not surprising to note that Fox has acquired a spectrum of vitri-
olic critics ranging from the chemist Stanley Miller to the crea-
tionist Duane Gish and the astronomer Carl Sagan. In fact, the
one point that the evolutionists and creationists both seem to
agree upon is the irrelevance of the work of Sidney Fox. When
such scientific eminences start getting hot under the collar, it’s
usually a sign that somebody’s doing something right. So let’s
dig a little deeper into Fox’s proteinoid idea and see why it
raises so many hackles.

As we’ve seen, amino acids can be readily produced in Miller-
style experiments. However, amino acids don’t easily unite to

PARADIGMS LOST

form peptides (short protein chains) in the presence of water. In
fact, just the opposite occurs: In water, peptides and proteins
break down into their amino acid constituents. The remedy
seems obvious: Just heat up dry amino acids so that the water
that’s formed when they join into a protein chain is carried off
as vapor. Oddly enough, when this experiment is carried out
with amino acids in the ratios found in naturally occurring pro-
teins, all that’s formed is a horrible, sticky, smelly, brown tar
instead of the prized protein chains. Enter Sidney Fox.

Rather than using the usual prescriptions for heating amino
acids, Fox found that different types of amino acids wouldn’t
hook together well unless extra amounts of any of three special
amino acids — lysine, aspartic acid, or glutamic acid — were pre-
sent. When these new mixtures were heated in the dry state at
temperatures up to 130°C, they rapidly formed polymer chains
of amino acids, but chains that didn’t correspond to proteins oc-
curring in earthly biology. For this reason, Fox termed these
products proteinoids.

Despite their unearthly nature, Fox’s proteinoids were found
to display certain features of interest. For instance, some of
them showed a catalytic capacity for several types of chemical
reactions, although the activity was not substantially better than
that shown by the same amino acid mixture before it was heated.
However, what was remarkable was the behavior demonstrated
by certain types of proteinoids when they were dissolved in
warm water and allowed to cool slowly. Under this very simple
operation, billions of microspheres formed from just a single
gram of proteinoid. Fox found that these microspheres would
grow and bud off smaller spheres, and that they had a somewhat
nonspecific enzymic activity; i.e., they would catalyze a fairly
broad range of chemical reactions. The “metabolism” of the pro-
teinoids is far less specific than that of Oparin’s coacervates, but
then Fox didn’t add any biological enzymes from the outside to
push the metabolism along. It’s vital to note, however, that just
like the coacervates, the proteinoids lack a hereditary mechanism
and will not evolve by natural selection. The main steps in Fox’s
road to life are displayed in the following diagram:

FOX'S SCENARIO

Amino acids -* Proteinoid chains -* Cells -* Genes

A WARM LITTLE POND

As noted, criticism of Fox’s proteinoid idea has been hot and
heavy ever since he first introduced the notion over three
decades ago. Many of the early complaints focused upon the geo-
logical question of where on the early Earth one would find the
sort of conditions needed to form the proteinoids. Stanley Miller
and Leslie Orgel ask whether there is any place on the present-
day Earth where all the necessary conditions are present, com-
ing to the sad conclusion that “we cannot think of a single such
place.” Earlier, Harold Urey stated quite unequivocally that “it
is difficult to see how the processes advocated by Fox could have
been important in the synthesis of organic compounds.” Re-
cently Fox has answered some of these geologically based diffi-
culties by noting that perhaps the proteinoids arose near the
thermal vents at the bottom of the Pacific Ocean. Somehow it’s
hard to see how the necessary dry heating could take place at the
bottom of the sea, but that’s the illogic of real science for you!
Other arguments against the proteinoids center upon the fact
that similar microspheres are created under a variety of circum-
stances, such as when ash forms out of molten lava in volcanic
explosions like the one that truncated Mount St. Helens. Yet
none of these microspheres show the capacity to grow, repro-
duce, and evolve in a manner that copies the internal organiza-
tion of the system. In other words, they don’t show a capacity
for self -organization, hence for life.

It’s clear that both the Oparin and the Fox scenarios are hope-
lessly deficient when it comes to the problem of providing a gen-
etic mechanism whereby hereditary information can be passed
along to future generations of cells, opening up the possibility
for natural selection to come into play. So just as the naked ge-
nies suffer from an Achilles’ left heel of no proteins to catalyze
reactions that would allow development of a large genetic infor-
mation store, the proteinists suffer from the complementary
right heel of no replication machinery. Since it seems difficult to
conceive of plausible ways to fill in the gaping holes in the argu-
ments of either the genies or the proteinists, perhaps the answer
lies in adopting a Hegelian dialectical stance, and attempting to
combine the best features of the two schools into a Dual-Origin
Theory (or Double-Origin Hypothesis). Such a theory would
argue that life emerged not once but twice, with the proteins and
replicators arising independently and then later linking up in a
mutually beneficial symbiotic arrangement. Let’s see how such a

100

PARADIGMS LOST

theory might work to plug the leaks in the all-or-none hopes and
dreams of the genies and proteinists.

LIFE: A TWICE-TOLD TALE

Some time ago at the Smithsonian Institution in Washington,
D.C., one of the more popular exhibits was a videotape showing
the famous TY chef Julia Child mixing up a batch of primordial
soup in vivid, living color. Unfortunately, Nature’s kitchen, just
like those of many of Julia’s TV fans, suffers from the unhappy
fact that knowing Julia’s methods and obtaining her results are
two very different matters. And as entertaining and educational
as the Smithsonian display was, the unvarnished truth is that
the delightful concoction coming out of Julia Child’s soup pot
had a flavor unlikely to have been on any earthly menu at the
dawning of life. Since this observation bears heavily upon the
Dual-Origin Theory of life, it’s worth our taking a moment to
scrutinize the recipe in greater detail.

A point of contact between the proteinist and the naked genie
programs for the origin of life is the assumption that all the
necessary raw materials could have been assembled by natural
means in the primordial soup. Miller-style experiments make
this assumption at least defensible for the proteinists, since
amino acids seem to form spontaneously in almost any kind of
primitive environment — just as long as it doesn’t contain any
appreciable amount of free oxygen. Thus, we can at least iden-
tify a path whereby simple proteins might naturally form. The
picture is far fuzzier for the natural assembly of nucleotides.

The bitter facts of chemical life are that it’s just plain hard to
see how nucleotides could have easily been made in the environ-
ment of the early Earth. For over three decades an army of
talented chemists has experimented endlessly with various for-
mulas for constructing nucleotides in the laboratory, with only
limited success. Some of the nucleotide bases have been created
from elementary compounds, but only under conditions that
would require a cold soup rather than the postulated hot primor-
dial soup. It has also proven possible to synthesize the sugar
components of nucleotides using formaldehyde, but again under
circumstances that are far more special than those needed for
creating amino acids via Miller-type experiments. Fortunately

A WARM LITTLE POND

101

the phosphate components of nucleotides don’t have to be syn-
thesized, since they occur naturally in rocks and seawater.

The major difficulty in nucleotide synthesis is in getting the
three components — bases, sugars, and phosphates — to stick to-
gether naturally in the right kind of geometrical arrangement.
If the linkages are made randomly, only about 1 percent of them
will turn out to be correct, and there appears to be no natural
mechanism that would be able to distinguish the one correct ar-
rangement from the other ninety-nine. In addition, nucleotides
are unstable in water and have the depressing tendency to dis-
solve back into their component parts. Thus, the rate of forma-
tion would have to be very high in order to counterbalance the
correspondingly high rate of decomposition in seawater. No one
has yet discovered a natural chemical mechanism that would en-
able nucleotides to be generated rapidly enough to find each
other and then form into the necessary double helices before
they fall apart by hydrolysis. This is one important fact favor-
ing a theory of life that requires only amino acids to be prebioti-
cally formed, with the nucleotides coming later as a by-product
of protein metabolism. But this is not the only argument that
speaks for the Double-Origin Hypothesis. Here are two more:

• Parasitism: Within the cellular cytoplasm (where the proteins
are manufactured), we find the organelles (mitochondria and
chloroplasts), which serve a vital function in extracting the en-
ergy needed for the cell to carry on its business. The organelles
have their own genetic machinery, which operates indepen-
dently of that found in the cell’s nucleus. As a result of stud-
ies of the cellular evolutionary tree, it’s been found that the
genetic apparatus of the organelles belongs to a different
branch of the tree than that present in the nuclei of eukaryotic
cells. The American biologist Lynn Margulis has forcefully
pressed the claim that this fact suggests that the organelles
originally lived a life totally independent of the eukaryotic
cells, and only later joined up with them in a parasitical, sym-
biotic relationship, probably to help the cell extract energy
from the environment more efficiently. It’s also been discov-
ered that the genetic code used by the mitochondria differs
slightly from that of the cell nuclei. Significantly, the differ-
ence is small enough to point to the conclusion that the two
codes must be related, tracing their origin to a common ances-

102

PARADIGMS LOST

tor. Both of these facts lend support to the Dual-Origin The-
ory. Of course, it should be noted that this is a situation in
which one DNA organism invaded another. It says nothing
about a “no-DNA” organism.

• Fossil evidence: In the oldest rocks that can be reliably dated
(about 3 billion years old), evidence is found of fossils that
bear a resemblance to modem bacteria, i.e., prokaryotic, sin-
gle-celled creatures. In rocks about a billion years younger, we
find traces of fossils similar to modern prokaryotic algae, in-
cluding multicellular entities. Finally, in rocks that are about
a billion years old, evidence of modern eukaryotic cells finally
appears. Unfortunately, the techniques available offer us no
way to determine whether or not the oldest fossils possessed a
modern genetic apparatus, or were cells with no nucleic acid
whatsoever. The only thing we can say with confidence is that
of the fossils originating over the past 1 billion years, all are
modern in form with contemporary eukaryotic features, in-
cluding genetic machinery. Thus the fossil record provides evi-
dence only for some sort of ancient living beings, but no
evidence at all that these organisms possessed any kind of rep-
lication apparatus utilizing nucleic acids.

Putting the chemical, fossil, and parasitism arguments to-
gether with the earlier difficulties in the arguments of the protei-
nists and the genies we come to a Double-Origin Theory in which
the first living agents were metabolizers (proteins), with the gen-
etic replication machinery following much later as a consequence
of the chemical reactions catalyzed by the primeval proteins. As
we’ve already seen, there’s no particular difficulty in forming
simple proteins via Miller-style reactions. So the essential step in
any kind of double-origin scenario is to offer a plausible means
by which protein replication can take place without invoking nu-
cleic acids. Robert Shapiro has suggested the following scheme
based upon the manner in which transfer RNA works in a mod-
ern cell.

Earlier we saw that when the ribosomes manufacture a pro-
tein chain, a central role is reserved for tRNA synthetases , spe-
cial enzymes that act as “interpreters” by having very specific
geometries at each of their ends. The geometry at one end fits
exactly one nucleotide triplet (codon), while the geometry at the

A WARM LITTLE POND

103

opposite end of the enzyme fits only the amino acid correspond-
ing to the anti codon of the codon on the other end. It is these
tRNA synthetases that do the real job of translating from the
language of the genes (nucleic acids) to the language of the pro-
teins (amino acids). Shapiro argues that perhaps the same sys-
tem could work, but in a simpler way, for direct protein
replication. It’s of considerable interest to note that this “trans-
lation” seems to represent a second kind of genetic code. At pre-
sent, the workings of this code within the tRNA are the subject
of feverish research activity. The interested reader is invited to
consult the “To Dig Deeper” section for citations to some of the
recent work on this second genetic code.

Shapiro’s basic idea is that the protein molecule that was to be
copied became attached to some support so that it could be dis-
tinguished from those molecules that were not to be copied. The
molecule could then somehow be turned on its support so that
each of its constituent amino acids was exposed to the ambient
environment. As each successive amino acid was exposed, a suit-
able interpreter enzyme would recognize the amino acid and
match it at its other end with exactly the same amino acid from
the environment, adding this new acid to a growing chain under
construction. Notice that this kind of matching requires a sim-
pler sort of interpreter enzyme than modern tRNA synthetase,
since the “protein interpreter” needs only to be able to recognize
the same kind of amino acid at both its ends. Thus it needs to
know only the language of proteins, not both the language of
proteins and that of nucleic acids. Assuming there ever was such
a system, at some stage it was eliminated in favor of the current
nucleic-acid-based method, implying that the protein replication
process was inaccurate, slow, inefficient, or defective in some
other way. However, the method does have the virtue of indicat-
ing how proteins could replicate themselves, as well as suggest-
ing why modern life uses only a few of the many possible amino
acids.

In the Miller experiment, the most prominent amino acids pre-
sent were the two simplest (in terms of number of atomic compo-
nents)— glycine and alanine. It’s reasonable to suppose that
these two would be present in the initial set of amino acids used
by the first proteins. On the other hand, the most complex amino
acids used in living forms cannot be produced even with the
most elaborate pre biotic simulations, and probably emerged

104 PARADIGMS LOST

much later as a result of earlier metabolic processes. Various
theoretical arguments have been produced showing that only a
handful of amino acids, say between four and six, are needed to
approximate the shapes seen in proteins today. The successive
introduction of each additional amino acid very likely repre-
sented a milestone in the evolutionary struggle of early protein
life, greatly increasing the “catalytic power” of the protein
chains that could then be formed. Ultimately a crossover point
was reached, where it required more work to create the copying
machinery for an additional amino acid than the effort bought in
extra catalytic power. At this point, natural selection would then
act to stabilize the menu of available amino acid components at
its current level of twenty. Would that we were so lucky with
the menu at the corner Chinese restaurant! Even granting the
above sequence of events as a starting point for life, how could
the nucleic acid replication process ever have gotten a foothold,
eventually to displace the protein replication apparatus!

In Shapiro’s setup, RNA and DNA arise only when phos-
phates become more readily available as rocks gradually erode
and dump more and more phosphates into the primordial sea.
The original nucleotide material consisting of sugars and phos-
phates would then be used as structural materials, as they are
even today in the ribosomes. One way that this original struc-
tural material could have been transformed into today’s genetic
apparatus would involve the development of short, specialized
units of RNA, each associated with a particular amino acid, as
well as the development of a longer RNA strand for every useful
protein. With this sort of innovation, the information present in
each protein would then also be stored in the RNA, giving a du-
plicate genetic system that would eventually be found more ef-
ficient at replication than the earlier protein-based system.
Finally, natural selection would ruthlessly assert itself and dis-
card the old replication apparatus in favor of the nucleic acid
system that we know today.

The reader should note that the above scheme avoids the buga-
boo of the genies, namely, establishing the means by which the
individual nucleotide subunits are assembled to form the first
strand of RNA. Recall that from the experiments of Eigen, Q/3
replicase can assemble a strand of RNA on its own, given a sup-
ply of the subunits. This step is very simple provided that the
replicase enzyme is already present. And there’s no problem in

A WARM LITTLE POND

105

envisioning how this enzyme might come to be available in a sce-
nario in which proteins came first, followed much later by RNA
and DNA. The Shapiro scenario given above suffers from the
same kind of assembly difficulty as the nucleic acid replicators of
the genies, i.e., how did the appropriate subunits come together
to form the first self-replicating system? However, the problem
is easier to rationalize and give plausible answers to using pro-
teins rather than nucleic acids as the first living forms.

Recently the physicist Freeman Dyson has proposed a quanti-
tative model for the Double-Origin Hypothesis, in which he ex-
plores the feasibility of the overall notion using what he terms a
“toy model” of the process of cellular metabolism. While there’s
no room here to enter into the details of Dyson’s model, it’s in-
teresting to examine one or two of his main conclusions. Follow-
ing the imposition of a variety of simplifying physical and
mathematical assumptions, the essence of Dyson’s model comes
down to the interrelationship of three parameters: a, a measure
related to the number of distinct amino acid building blocks
(technically, monomers ) composing the original living objects; b,
a measure of the number of distinct sorts of chemical reactions
that the primitive life forms were capable of catalyzing; and N,
the size of the molecular population in a chain composing such a
form. What Dyson is interested in is those combinations of a,b,
and N that allow a reasonable possibility for the system to jump
from a disordered state of miscellaneous chemicals to the or-
dered state of a living agent.

In analyzing the consequences of his model, Dyson discovered
that the only values of the parameters that resulted in physi-
cally interesting behavior were those in the ranges:

a: from 8 to 10

b: from 60 to 100

N: from 2,000 to 20,000

When translated back to physical units, this result implies that
the number of monomer types should range from nine to eleven.
As we know, in modern proteins there are twenty types of amino
acid monomers, so it’s reasonable to suppose that ten or so would
be enough to provide sufficient diversity of protein function to
get life off to a start. At the other end, the model definitely fails

106

PARADIGMS LOST

if a =3. This implies that life according to Dyson could not
possibly have begun with only the four nucleotides forming
modern RNA; nucleotides alone just don’t offer great enough
chemical diversity to make the transition from disorder to order.
Thus, the model displays a pronounced bias in favor of proteins
as opposed to nucleic acids as the material basis of life.

Having the discrimination factor b in the range from 60 to
100 turns out to be chemically reasonable for the first primitive
proteins, and also endows the model with the all-important prop-
erty of being able to tolerate very high error rates. If one were
to assume exact replication from the very beginning with a
low tolerance of errors, the jump of a chain of N monomers
from disorder to order will occur with a probability of around
(1 + a )-N. This implies that a replicating system can spontane-
ously emerge only if A- is no greater than about 100, as noted in
an earlier section. However, in Dyson’s nonreplicating system
with a and b in the ranges above, the error rate will be about 25
to 30 percent, and still a chain of ten thousand or more mono-
mers can make the transition from a disordered state to an or-
dered one with reasonably high likelihood. Such a level of per-
formance in which only three out of every four links in the chain
are correctly placed would be intolerable for a replicating sys-
tem, but is quite acceptable for a nonreplicating one.

The overall behavior of Dyson’s model is summarized in Fig-
ure 2.7, in which each point corresponds to a particular choice of
a and b. Models that admit the possibility of both ordered and
disordered states occupy the central region in the diagram la-
beled the “transition region.” The biologically interesting mod-
els are those near the cusp, which have high error rates and are
able to make the transition from disorder to order with large
population sizes. One interesting case discussed in detail by
Dyson is when a = 8, b = 64, leading to an error rate of exactly
one third and a critical population value of Nc = 26,566. The
region labeled “dead” in Figure 2.7 corresponds to models that
have only a disordered state. Such models have a too large (too
much chemical diversity) and b too small (too weak catalytic ac-
tivity) to produce an ordered state. Conversely, the region la-
beled “immortal” has a too small (too little chemical diversity)
and b too large (too strong catalytic activity) to produce a disor-
dered state. In further discussion of this model, Dyson also
gives some provocative arguments for how the asymmetry be-

A WARM LITTLE POND

107

FIGURE 2.7 Summary of Dyson’s model

tween life and death could arise from such a system, and why it
is that death is so much easier than resurrection. But my sys-
tem-theoretic prejudices have probably already caused me to de-
vote too much attention this model, so I’ll let those readers with
a hankering for rising from the dead consult “To Dig Deeper”
and pass now to a summary of the Dual-Origin Theory.

All but the most casual of readers will have long ago realized
that there’s really not such a great difference between the Dual-
Origin Theories of Shapiro and Dyson and the proteinist argu-
ments of Oparin and Pox. The main point of departure is that
both Oparin and Fox argue that the replication machinery came
about early in the game and, moreover, arose directly out of the
initial metabolism. The “doublets” argue that the genetic appa-
ratus was a Johnny-come-lately on the origins scene, and did not
arise directly out of the initial proteins but rather had a quite
different structural function originally, and that its ultimate
role as a replicator arose as a type of “genetic takeover” from

108

PARADIGMS LOST

the original protein replication mechanism. Diagrammatically,
we have:

THE SH API RO-DYSON SCENARIO

Cells -*■ Proteins -» — RNA -» DNA
much later

The most striking aspect of the doublet claim is that it goes
straight in the face of the cherished Central Dogma of Molecular
Biology, discussed earlier. Should such a pillar of modern bio-
logical wisdom be so lightly discarded? Well, as we noted earlier,
the originator of the dogma, Francis Crick himself, has stated
that not only did he misunderstand the meaning of the term, but
he also meant for the principle to apply only to modern orga-
nisms; he makes no claims for how ancient organisms might have
functioned. A remark made by one of the most prominent genies,
Leslie Orgel, also bears upon this point. When Robert Shapiro
outlined his doublet theory to him, Orgel replied, “Enzymes can
do anything!” By this he meant that enzymes could, in principle,
carry out the replication functions suggested in Shapiro’s the-
ory. But that doesn’t prove that such a scheme ever existed.
What’s needed is a plausible physical mechanism by which the
protein replication process could have gotten started. Mainline
doublets like Shapiro and Dyson think this could have happened
using the carbon-based compounds composing modern life. The
Scottish chemist A. G. Cairns-Smith says that carbon is much
too high tech a material for this job, and has offered a fascinat-
ing silicon-based alternative with the claim that, just as the
Bible says, life started as a mere mote of dust in someone’s eye.
We devote the next section to an account of these ideas.

ASHES TO ASHES, LIFE FROM DUST

One of the multitude of ways I managed to misspend my youth
in the 1950s was by hanging around the local cinemas. Rather
than devoting valuable energy to homework, practicing the
piano, or some other dull character-building task, I squandered
my time (in my mother’s opinion, at least) in soaking up the
cinematic offerings of the day at what I then saw as the near

A WARM LITTLE POND

109

budget-breaking admission price of one dollar. In order to put
as much distance as possible between myself and the onerous
program of chores lined up for me by my teachers and parents, I
always tried to arrange to spend my weekly cinema allowance on
triple features, which in those days generally meant an entire
afternoon of the then-popular science-fiction and horror films.
One of the films that I still recall fondly was the 1957 classic The
Monolith Monsters, a story about some sort of silicon-based life
forms that were transported to Earth on a meteorite. These
strange objects somehow absorbed silicon from Earth’s sand and
rocks, using it to grow into tall monoliths that eventually top-
pled over and broke up into small pieces, which then started
growing again into even more of the “rock monsters,” whose
“life cycle” is shown in Figure 2.8. The film’s crescendo was
reached as a forest of these monoliths threatened to pulverize
some rural community in the Arizona desert. They were stopped
an eyelash away from town only when the film’s brilliant and
handsome scientist-hero realized that the monsters’ growth could
be stopped dead in its tracks by the salt in simple seawater.
While pretty farfetched in regard to both its science and its sci-
entists, The Monolith Monsters represented an entertaining Hol-
lywood attempt to speculate on the nature of life forms based
upon silicon, an idea that has recently been resurrected by A. G.
Caims-Smith as the material basis for his Dual-Origin Theory
of life.

The motivation behind Cairns-Smith’s revival of silicon is that
what’s important in any origins theory is to find some system
that will get metabolism and replication started. He conjectures
that a “low-tech,” silicon-based setup might be easier to get roll-
ing than the sort of high-tech, carbon-based systems discussed so
far. Once some kind of living system was up and running,
Cairns-Smith argues, the more efficient carbon-based units could
then engage in a “genetic takeover,” pushing the original system
out of the spotlight and back into the wings.

In his writings, Cairns-Smith identifies what he terms “seven
clues to the origin of life” as evidence to support his claims that
modern life got its start as a bunch of mud. The clues he cites are:

1. Biology: Genetic information is pure form, not substance, and

evolution can begin only when this kind of replicable form

exists.

110

PARADIGMS LOST

FIGURE 2.8 Crystalline growth in The Monolith Monsters

2. Biochemistry: DNA and RNA are biochemically complex and
hard-to-make molecules, suggesting that they were late arriv-
als on the evolutionary scale of things.

3. Construction industry: In evolution, things can be subtracted
as well as added. This can lead to the kind of mutual depen-
dence of components seen in the central biochemical pathways
of life.

4. Structure of ropes: Gene fibers, like rope fibers, may be added
and subtracted without breaking the overall continuity of the
gene line. This suggests how organisms based on one genetic
material could gradually evolve into organisms based on an
entirely different genetic material.

5. History of technology: Primitive machinery is usually different
in design and construction from later advanced counterparts.
The primitive machine has to be easy to make from immedi-
ately available materials, and must work with a minimum of
fuss. The advanced machine does not have to be particularly
easy to assemble, nor does it have to be made from simple
parts. This fact suggests that the first organisms would prob-
ably have been very different from the “high-tech” organisms
of today.

6. Chemistry: Crystals put themselves together in a way that
could be suitable for “low-tech” genetic materials, suggesting
a direction in which to look for the primitive biochemical ma-
terials.

7. Geology: There is a lot of clay continually being made through
natural processes. This kind of inorganic crystal seems to be

A WARM LITTLE POND

111

much more appropriate than big organic molecules for primi-
tive genes, as well as for other primitive control structures
like low- tech catalysts and membranes.

These clues from many different areas of science and technol-
ogy constitute tantalizing arguments for taking crystals of clay
seriously as the first living material. While there’s no room here
to go into the details underlying Cairns-Smith’s argument, it’s
definitely worthwhile to examine the general scenario he offers
for how events might plausibly have taken place. But in order to
make sense out of the steps in the Cairns-Smith theory, we first
need to understand just a bit about the basic physical properties
of clay and crystals.

Occasionally, despite careful planning, I’m forced into trying
to recall the odd fact or two from my late and unlamented high-
school and university chemistry courses. Usually the first thing
that comes to mind (following a prodigious effort) is an image of
the kind of experiment that everyone does along about the sec-
ond week of such a course: the creation of crystals of salt from a
supersaturated solution of sodium and chlorine ions. The experi-
ment generally involves dumping an enormous amount of salt
into a beaker of water, heating the water to boiling in order to
dissolve the salt into its component ions, and then letting the
resultant liquid slowly cool, being very careful not to agitate the
container. After the liquid mixture cools down, a tiny crystal of
salt is dropped into the flask as a seed to start the crystallization
process. Almost immediately the ions dissolved in the water start
to attach themselves to the seed, forming long crystals of salt
that eventually break up, the pieces serving as seeds for further
crystallization until the dissolved salt is exhausted. The basic
process taking place is illustrated in Figure 2.9, where the black
dots represent atoms of sodium and the white dots are chlorine.
While not all crystals are as simple as salt, many of them having
more exotic geometries than a cube and many of them containing
repetitions of several layers composed of more than two atomic
ions, the salt crystal depicted in Figure 2.9 illustrates several
properties of crystals crucial for Cairns-Smith’s theory.

First of all, note that the crystalline structure is very regular,
consisting of a repetitious pattern of a two-dimensional lattice
structure at each of whose points there is either a sodium or a

112

PARADIGMS LOST

O-

FIGURE 2.9 Crystallization of salt from sodium and chlorine

chlorine atom. This kind of pattern gives structural integrity to
the crystal, as well as providing it with the opportunity to grow
by adding additional atoms from the surrounding environment,
as shown in the figure. In many crystals the atomic bonds be-
tween the various layers are rather weak, much weaker in fact
than those holding the atoms together within the layer. Conse-
quently, it’s very easy for layers to shear along the natural crys-
talline planes, just as the salt crystals in the experiment break
up when they get too big. The combination of being able to at-
tract new ions and shearing along natural planes of cleavage
means that crystals can certainly grow and multiply, i.e., display
a kind of “metabolism.” What would be needed for life is some
means by which the crystals could evolve.

To see how crystalline growth could take place with variation
in a self-propagating manner, we have to depart from the
foregoing fiction of idealized crystal growth as presented in ele-
mentary courses, and consider how crystals really grow in na-
ture. In real life, crystals are not the perfect, infinite lattice
structures depicted in Figure 2.9 or school textbooks, but rather
grow with defects of both a mechanical and a chemical nature.
As seen in Figure 2.8, many crystals contain notches, grooves,
and other types of mechanical flaws that can be passed along,
layer by layer, as the crystal is formed. Other types of mechani-
cal defects come about when growth rates differ on the various
faces of the crystal, leading to separate “domains” that are
slightly out of alignment but each still growing according to the

A WARM LITTLE POND

113

overall plan. Now suppose that two identical crystals start grow-
ing in the same environment, but that soon one of them starts
acquiring variations of the above type. Further, imagine that
this variation in physical shape somehow allows this “mutant”
form to replicate faster than the other, perhaps by being re-
tained more effectively within the pores of a rock from which the
other form is more easily washed away. By natural selection,
this more “efficient” form may soon come to dominate the crys-
tal population.

But such mechanical imperfections are not the only way that
crystals can evolve. Each layer composing a crystal contains
many atoms, one at each lattice site. For instance, most clays
consist of layers of oxygen ions with layers of positively charged
ions (usually silicon or aluminum) sandwiched in between. In
many such clays, one of these positive ions can be replaced by
another type without destroying the clay’s capacity to grow.
Such substitution patterns can become quite intricate, making
the surface pattern of the clay a very complex chemical struc-
ture. Furthermore, the pattern can be passed along (inherited)
by subsequent layers as they’re added. This inheritance can be
by a direct matching of the pattern, or it may involve the forma-
tion of some sort of “complementary” layer. Of course, such a
complementary matching process would be very close to what we
see in DNA transcription and replication. Thus we see the possi-
bility for “crystal genes,” by both mechanical and chemical
means, to form and act to perpetuate the information needed to
give rise to further generations of crystalline forms. But is this
enough to constitute life? And how and why do modern organic
proteins and nucleic acids enter the picture?

Cairns-Smith’s answer to these vital questions is radically
different from the conventional tales told by the genies, protein-
ists, and doublets. His claim is that “living” crystals started
making the organic components of modern life in order to help
themselves survive and multiply. While there’s no room here to
enter into the intricate arguments he makes to support this
claim, the overall conclusion is that it’s at least plausible that
the manufacture of organic compounds could help the clays in
several ways: by providing mechanical support, by removing un-
desirable ions, by controlling the size and structure of the crys-
tals, by assisting in the capture of inorganic ions, and so on.
Now what about the takeover? When and why did the original

114

PARADIGMS LOST

crystalline life get shoved aside, to be replaced by its “assist-
ants,” the carbon-based units?

According to Cairns-Smith, the takeover by organic life took
place when some organic forms within the crystal life began to
reproduce themselves at a rate that exceeded that of their crys-
talline hosts. At this very moment, the crystals’ days were num-
bered as the dominant life forms on Earth. Once crystals had
made the first strand of self -replicating (and possibly self -cat-
alyzing) RNA, they would have created a much more efficient
and versatile genetic material than their own “low-tech” genes.
Natural selection would then have seen to it that the “high-tech”
nucleic-acid-based genes moved into the spotlight, eventually
pensioning off the crystal life as museum pieces. We can summa-
rize the Cairns-Smith program for the unfolding of life in the
following diagram:

THE CAIRNS-SMITH SCENARIO

Clay -* Growth/Replication -* Organics -» RNA/Proteins

takeover

The great advantage of Cairns-Smith’s Clay Theory is that it
makes it far easier to see how life could have gotten started.
There was no need for an unlikely and fortuitous juxtaposition
of chemical and geological events, only simple chemical reactions
involving readily available materials; in fact, these reactions are
still going on today. But if such chemical activities are still hap-
pening, and the sequence of events is so easy, why don’t we see
any crystal life today? Or do we?

Currently no one really knows the answer to these queries,
principally because no one’s really looked. However, in his books
and articles Cairns-Smith suggests a number of places where we
might look for evidence of such life and what form we might
expect it to take. For instance, he suggests that if such crystal-
line life exists today it might simply be a rather loose collection
of interacting crystals whose boundaries are somewhat fuzzy
and diffuse. Consequently, we should look for “bizarre” crystal
structures doing unusual things.

Cairns-Smith also suggests laboratory experiments involving
a mineral version of Spiegelman’s test-tube evolution experi-
ment with the Qy3 virus. A supersaturated solution of minerals

A WARM LITTLE POND

115

would flow into a continuous crystallizer. Crystal formation and
growth would take place within it, with a suspension of crystals
flowing out the other end. Imagine that two different kinds of
crystals form in the crystallizer, and that one grows quickly but
doesn’t break up and so is eventually washed out the other end.
But suppose the second type not only grows, but easily breaks
apart, and that new crystals are formed to compensate for those
lost at the exit pipe. If some random variant can replicate itself
more rapidly than the competition, it should eventually take
over the entire crystallizer, just as the “monster” did in Spiegel-
man’s test tube. This kind of experiment has yet to be done, but
would shed considerable light on the possibilities of crystal evo-
lution and hence on the feasibility of Cairns-Smith’s theory of
life.

With the Clay Theory, the Prosecution has completed its case
for the origin of life on Earth via natural chemical and geologi-
cal processes. Now we ask the Defense to take the floor and pre-
sent its spectrum of otherworldly claims for how life got its
start. The case for the Defense rests on two pillars: arguments
from Nature and arguments from the supernatural. We will lis-
ten to natural claims first.

IT CAME FROM OUTER SPACE

James Watson opens The Double Helix, his classic account of
science in the fast lane, with the statement that “I have never
seen Francis Crick in a modest mood.” Crick is now a seventy-
ish, graying, distinguished-looking man of good cheer, who ap-
pears to have mellowed considerably since Watson’s account.
But whether you term it good cheer, immodesty, or just plain
brash exuberance, Francis Crick has spent the better part of the
last three decades on the front pages of both the scientific and
popular press, with a steady succession of offbeat and slightly
outrageous ideas about various aspects of molecular biology,
brain theory, extraterrestrials, and other matters of body and
mind.

One of Crick’s more speculative offerings appeared in a 1973
article coauthored with Leslie Orgel, in which they claimed that
life on Earth might have originated in outer space with extrater-
restrials. Crick later expanded his ET theory into the book Life

116

PARADIGMS LOST

Itself, which argues that the Earth has been under continuous
observation by intelligent extraterrestrials who, when the time
was right, planted the “seeds” of life on Earth. This notion, like
Linus Pauling’s theory of vitamin C and the common cold,
would probably sink like a stone into the seas of scientific obliv-
ion if it were not being championed by a Nobel laureate. Never-
theless, given the nontrivial obstacles in the path of all of the
Earth-based scenarios for life, it’s definitely worth taking a look
at what Crick has in mind.

The principle underlying Crick’s “life from space” thesis had
its origins early in this century with the ideas of Svante Arr-
henius, who promoted a vision of life raining down on Earth in
the form of tiny spores from space. Arrhenius was a Swedish
chemist whose original work on the behavior of salts when they
dissolve in water was so lightly regarded that when presented in
support of his Ph.D. degree, it received the lowest possible pass-
ing mark. Later, in the sort of 180-degree turnabout that scien-
tific cranks dream of, his ideas were vindicated and he was
finally rewarded with the Nobel Prize for chemistry in 1903. His
scientific position secure, Arrhenius could then afford the luxury
of proposing a theory of life that involved microorganisms es-
caping from other life-bearing planets in the galaxy, traveling
across interstellar space propelled by the pressure of stellar ra-
diation. According to Arrhenius, one of these spores eventually
landed on Earth, giving rise to life as we know it today. This
Panspermia Theory is now mostly discredited, the principal ar-
guments against it being that it’s unlikely that even one such
spore would arrive on Earth during the entire history of the
universe, and furthermore, any microorganism of the type we
know today would very likely have been killed by solar radiation
and/or the cold and vacuum of space.

Crick updated the Panspermia Theory by noting that most of
the arguments against it would be invalid if the spores arrived
on Earth after having been transported here on some kind of
interplanetary vehicle. Observing that the universe is more than
twice as old as the Earth, Crick argued that it’s not unreason-
able to suppose that life could have arisen more than once. Fur-
thermore, he rightly pointed out that there’s no reason to believe
that the conditions that prevailed here on Earth were anywhere
near optimal for the development of life. Putting these remarks
together with the anthropomorphic hypothesis that any extrater-

A WARM LITTLE POND

117

restrial life form would have the same psychological need for
expansion that humans have displayed, he concluded that the
most likely explanation for life on Earth is that it was seeded by
extraterrestrials .

While “Directed Panspermia” formally answers the question
of how life arose on Earth, from the standpoint of a scientific
explanation it does so in the most unsatisfactory way imagin-
able— by pushing the problem off into some other solar system.
In his defense, Crick himself appears not to take the whole busi-
ness very seriously, and has commented that he put the hypothe-
sis forward only to focus public attention more sharply upon the
difficulties associated with the origin-of-life question. In fact,
even Crick’s own wife thought he’d gone slightly mad, dismiss-
ing the whole notion as pure science fiction. But in contrast to
the playful ET origins suggested by Crick, another eminent
British scientist, Sir Fred Hoyle, has put forth a different sort
of “life from space” theory, one that he takes very seriously in-
deed.

Is there something about the air of the British Isles that
causes responsible, rational, sober scientists to turn their atten-
tions to eccentric, crankish, or just plain weird notions when
they begin to enter their philosophizing years? Isaac Newton ap-
pears to have caught this disease and spent most of his later
years hunched over the Bible in search of ammunition to sup-
port his claims for familial relationships that apparently only he
could see. The case of Bertrand Russell’s offbeat social theories
is well chronicled, and Francis Crick’s Directed Panspermia
seems also to represent a mild dose of this peculiarly British
affliction. Fred Hoyle, on the other hand, appears to many to
have caught a terminal case with his advocacy of the idea that
life on Earth originated as a kind of “disease” from the stars.

Hoyle, a rather short, vigorous man in his mid-seventies, who
still tirelessly stalks the moors of his native Yorkshire, has had a
long and distinguished scientific career noted for pioneering
work involving the way in which heavier elements are made from
lighter ones in the interiors of stars. He is also known for his
now somewhat discredited theory of the so-called steady-state
universe, formulated together with Thomas Gold of Cornell, a
theory in which the universe didn’t begin with a bang, but
rather has always been more or less the same whimper we see

118

PARADIGMS LOST

today. For these universally acknowledged scientific contribu-
tions, Hoyle has been recognized with election both as a fellow of
the Royal Society and as a foreign associate of the U.S. National
Academy of Sciences, as well as having been honored with a
knighthood in 1972. Besides his real science, Hoyle has also
found time to make major contributions to the literature of sci-
ence fiction, having written several very intriguing and enter-
taining novels including one of the all-time classics, The Black
Cloud, which introduced the possibility of an alien life form
composed of a gigantic cloud of interstellar plasma.

With his penchant for pushing contentious scientific causes,
it’s perhaps not surprising to learn that Hoyle has also found
time to get into all sorts of squabbles over academic, political,
and administrative matters with his colleagues, especially those
at Cambridge, his home university. At one time in the mid-
1960s, the heat was so intense that Hoyle resigned his position
on the mathematics faculty, threatening to emigrate to the
United States. This fate was staved off only by his appointment
as director of the newly formed Institute of Theoretical Astron-
omy. Somewhat later, Hoyle also found his way into the public
press when he accused his colleague Anthony Hewish of exploit-
ing the work of a graduate student, Jocelyn Bell, in promoting
the work that eventually led to the Nobel Prize for physics in
1974 for the discovery of pulsars, as recounted in the last chap-
ter. With such a track record, the appearance of Hoyle and his
ideas in polite scientific circles is about as welcome as the ap-
pearance of Martin Bormann at a bar mitzvah. Nevertheless,
Hoyle and his longtime associate Chandra Wickramasinghe have
put forth not just one, but two distinct scenarios for how life
came to be. Let’s briefly look at these two visions of life accord-
ing to Hoyle.

H&W: VERSION I

In their early papers, Hoyle and Wickramasinghe claimed that
life originated in the molecular clouds of interstellar space, and
was then transported to Earth by comets. Radio astronomers
have noted that many of the important organic molecules needed
for life are present as major components of the vast clouds wan-
dering between the stars. H&W jumped on this fact, claiming
that cometary material “seeded” the primordial soup with the
right stuff to develop into the first terrestrial life forms.

A WARM LITTLE POND

119

The essential point of this cometary theory was the assertion
by H&W that the interstellar dust grains that seeded the Earth
were grains of cellulose, perhaps the most abundant biological
product on Earth, forming the main component of trees, cotton,
and many other important plants. This claim raised a number of
scientific eyebrows, principally because cellulose is such a special
material, coming about on Earth only under very particular bio-
logical circumstances. Thus, any chemical process going on in
outer space that could yield such a specific substance could also
be expected to give a large number of other important chemical
products as well. This kind of claim seems so miraculous to the
flinty-eyed community of astrochemists that the most over-
whelming evidence and documentation would be required to sub-
stantiate it. Unfortunately, H&W offered no such weight of
evidence, proposing instead to back up their assertions only with
some rather inconsistent infrared spectral observations of dubi-
ous pedigree. This evidence involved averaging the spectral char-
acteristics of a collection of 153 compounds they thought were
relevant to life, and then smoothing the result to fit the observed
spectral data of the interstellar clouds. Spectroscopists and as-
trochemists around the world were uniform in their denuncia-
tion of this procedure.

When Hoyle and Wickramasinghe published their ideas Ur
the general public in the popular book Lifecloud, the response of
the scientific community was mixed in the extreme. At one end
of the spectrum were the scathing remarks by Lynn Margulis,
who called the book “flamboyantly irresponsible,” and noted
that “its theme moreover is entirely contrary to the considered
opinion of most workers in the field. . . .” On the other hand,
some words of praise came from science journalist John Grib-
bin, who in Genesis, his own book on the origin of life, states
that “something along these lines will eventually become the es-
tablished view.” Similar testimonials were forthcoming from
others as well, who noted that it should definitely be possible for
complex molecules to form spontaneously in cosmic gas clouds.
However, extensive scrutiny of the scientific arguments pre-
sented by H&W in the technical papers supporting their claims
uncovered so many holes that even the Watergate plumbers
couldn’t have made the Lifecloud Theory respectable. Besides
the aforementioned problems with the spectral data, H&W were
taken to task by their critics for a plethora of experimental
gaffes, ignorance of discontinuing data, statistical snafus, and,

120

PARADIGMS LOST

in general, “fingerprints” of the sort we discussed in Chapter
One under the heading of pseudoscience. Thus ended H&W’s
initial foray into the origin-of-life game; however, it was not to
be their last.

H&W: VERSION II

Following a half-time breather, H&W reentered the game with a
second theory that stands in almost total opposition to their
original ideas about interstellar dust and cometary messengers.
In Version II, all pretense at a natural explanation for life is
abandoned in favor of the claim that life originated with a Crea-
tor, and was then carried to Earth as a form of cosmic “dis-
ease.” As an indicator of the kind of about-face that the Disease
Theory represents, in the Lifecloud Theory H&W accepted the
primordial soup as the breeding ground for the life deposited by
the comets. But in their new vision, H&W state that “another
fuddled notion is that life began here on Earth in a thin brew of
organic material. The mystery is why grown men and women
have allowed themselves to be persuaded into such beliefs, in
spite of there being a considerable body of fact running against
them.”

Besides repudiating most of their original claims, H&W
brought many new and wonderous notions into the Disease The-
ory of life. For instance, they assert that many historical devel-
opments were caused by diseases originally brought here from
space. As an illustration, they cite the superiority of classical
armies to medieval ones, with the explanation that the Middle
Ages were riddled with diseases. This claim is then followed by
the even-more-difficult-to-swallow statement that “we also attri-
bute the rise of Christianity to the same disease-filled epoch.”

As noted, Hoyle and Wickramasinghe’s second theory attrib-
utes the origin of life to a Creator, but not just one of the deities
claimed by conventional religions. No, their Creator is one of
their own divination, being none other than — a silicon chipl Ap-
parently, they think that the dust clouds of space somehow coa-
lesced into such a chip, much the same way that the sentient
cloud in Hoyle’s famous science-fiction story The Black Cloud
was formed. Unfortunately they offer no scientific arguments or
testable predictions, nor do they cite any experimental data in
support of these strange notions. As H&W pile one extrava-

A WARM LITTLE POND 121

gance upon another, they ultimately wind up defending a posi-
tion that is unabashedly of the divine revelation variety.

Amusingly, Robert Shapiro has also noted the similarity of
the later H&W theory with the thesis put forth as science fiction
in The Black Cloud. The Cloud is surprised to find life on Earth,
stating that space is a far better place for the assembly of bio-
chemicals. Sensing higher intelligences in the universe, the
Cloud finally becomes bored with humans and sets off to find
these higher intellectual forms. Only by reading this early
(1957) fictional account does one realize that in Lifecloud and
Diseases from Space Hoyle finally came out of the closet to dis-
play a long-held, essentially religious view of the mystical ori-
gins of life on Earth. The only difference is that the intervening
decade has seen Hoyle’s vision pass from the realm of fiction to
that of “fact.” Thus do Hoyle and Wickramasinghe move from
scientific arguments for the origin of life to what are essentially
religious ones, treading exactly the same path as our next extra-
terrestrial-origins adherents, the “creationists,” only in pre-
cisely the opposite direction.

AND GOD CREATED. ..FROM FISH TO GISH

In an attempt to effect legislative repair to one of the oldest
flaws in the fabric of Nature, the state of Indiana in 1897
enacted a law setting the legal value of n at precisely 4, replac-
ing its inconvenient “natural,” but irrational, value tt =
3.14159265 . . . Later, a Tennessee legislator suggested the value
be legally fixed at 3, but this idea was immediately quashed when
a British clergyman, in one of those hilarious letters that Brit-
ish clergymen have traditionally sent to The Times of London,
stuck up for the Indiana value, stating that 3 was inadequate
since it wasn’t even an even number! But the Tennessee legisla-
ture eventually imposed its will on an unruly cosmos anyway by
enacting a different law making it illegal to teach evolution in
the classroom, an action thrusting the tiny hamlet of Dayton
into the international spotlight in 1925 with the celebrated Mon-
key Trial of John Scopes, a substitute for the local high-school
biology teacher, accused of filling the heads of his charges with
pernicious Darwinian visions.

For most of us, I suppose, the dramatic account of the Scopes

122

PARADIGMS LOST

trial in the film Inherit the Wind, in which a legendary barrister
based on Clarence Darrow (played by Spencer Tracy) crushes
the fundamentalist arguments of a prosecuting attorney mod-
eled on William Jennings Bryan (played by Predric March),
represented what we thought of as the death knell of legislative
tampering with Nature. This despite the fact that Scopes was
actually found guilty and assessed a one-hundred-dollar fine (al-
though two years later the Tennessee Supreme Court overturned
the conviction on technical grounds). And a death knell it was,
at least insofar as brute-force, frontal legislative assaults on Na-
ture by religious fundamentalists are concerned. But in March
1981, not to be outdone by its next-door neighbor, the Arkansas
state legislature revived the spirit of Dayton by resurrecting a
fundamentalist interpretation of the origin of life under the new
rubric “creation science.” With the enactment of the Balanced
Treatment for Creation Science and Evolution Science Act (Ar-
kansas Act 590), stating that “public schools in this state shall
give balanced treatment to creation science and to evolution sci-
ence,” the battle was rejoined between the fundamentalists and
the scientists, only this time it was to be fought on the home
ground of science rather than in the pulpits. Let’s take a mo-
ment to understand why.

The essential components of the “creationist” vision of the or-
igin of the the Earth and its life forms is contained in the fol-
lowing pledge sworn to by each member of the Creation
Research Society:

1. The Bible is the written Word of God, and because we believe
it to be inspired throughout, all of its assertions are histori-
cally and scientifically true in all the original autographs. To
the students of nature, this means that the account of origins
in Genesis is a factual presentation of simple historical
truths.

2. All basic types of living things, including man, were made by
direct creative acts of God during Creation Week as de-
scribed in Genesis. "Whatever biological changes have occur-
red since Creation have accomplished only changes within the
original created kinds.

In addition to swearing this pledge of “allegiance,” all prospec-
tive members of the society are also required to possess an ad-
vanced university degree in some field of science. As a result,

A WARM LITTLE POND

123

members in essence agree to forsake the common practices of
their profession in certain areas, and instead accept explana-
tions on the basis of divine authority alone.

In 1968 the U.S. Supreme Court outlawed all anti-evolution
laws like the Tennessee statute on the grounds that they violated
the constitutional prohibition against mixing the state, in the
form of the schools, with religion. Since this decision effectively
prevented the creationists from having their ideas of religion in-
troduced into the educational curricula, the fundamentalist
movement decided to settle for the next best thing and mounted
a campaign to push its position into the classrooms, dressing it
up as science. The Arkansas bill gives a particularly graphic ac-
count of the strategy employed. Arkansas Act 590 lists six prin-

(ciples of “evolution science” side by side with corresponding
principles of “creation science,” and then goes on to state that
both should be given equal time in the classrooms. The two most
important principles for our purposes are the following, which I
have taken directly from the text of the act: “Creation science
means the scientific evidences and related inferences that indi-
cate: (1) Sudden creation of the universe, energy, and life from
nothing; . . . (6) A relatively recent inception of the earth and
living kinds.” Other points of the act involve the occurrence of a
global flood, separate ancestry for man and apes, and other simi-
lar biblical stipulations. It’s clear from the above statements
that in order to make their case, the creationists are going to
have to attack the conventional scientific views on several as-
pects of geology, most importantly the matter of the age of the
Earth.

In speaking of the education of their children, creationists are
fond of citing the remark of William Jennings Bryan that
“Christians desire that their children shall be taught all the
sciences, but they do not want them to lose sight of the Rock of
Ages while they study the age of rocks.” This well-known re-
mark served for years as a rallying cry for fundamentalists as-
serting that the rocks of the Earth were only a few thousand
years old, just as claimed in Genesis. It doesn’t take too much
imagination to envision the loathing with which the creationists
look upon the increasingly accurate radiocarbon-dating methods
developed over the past few decades. With these unassailable
methods, used recently, for instance, to demonstrate the medie-

124

PARADIGMS LOST

val origin of the Shroud of Turin, the high levels of uncertainty
arising from the old fossil and sediment dating schemes were
eliminated, showing the Earth to be at least 4 billion years old.

How did the creationists react to such incontrovertible evi-
dence of an ancient Earth? Well, let me quote Henry Morris, a
hydraulics engineer and director of the Creation Research Soci-
ety: “The only way we can determine the true age of the earth is
for God to tell us what it is. And since he has told us, very
plainly, in the Holy Scriptures that it is several thousand years
in age, and no more, that ought to settle all basic questions of
chronology.” Such an act of faith unfortunately rejects data,
methods, experimental equipment, and all of the other parapher-
nalia of science. In fact, the leading creationists have been even
more candid in their rejection of science’s traditional methods of
inquiry.

Duane Gish holds a Ph.D. in biochemistry from the University
of California at Berkeley; he is also the vice-director of the Cre-
ation Research Society and a regular participant at university
debates on the merits of creation science. Since he is trained in the
scientific method, especially in an experimental science like bio-
chemistry, it’s odd, to say the least, to read in his book Evolution:
The Fossils Say No that “we do not know how the Creator cre-
ated, what processes He used, for He used processes which are
not now operating anywhere in the natural universe. . . . We
cannot discover by scientific investigation anything about the
creative processes used by the Creator.” With such statements,
creation “science” joins the long list of other perverse modern
“sciences,” such as “fashion science,” “dairy science,” and
“educational science,” all of which can be conveniently sub-
sumed under the heading “nonscientific science.”

Despite the cursory nature of our airing of the creationist
views, I think most readers will find no difficulty in understand-
ing the opinion of Judge William Overton in his ruling de-
claring the Arkansas Act 590 unconstitutional. Citing the
creationists’ own words in deciding that creation science was not
science but religion, the good judge offered one of the most con-
cise, best-thought-out lists of criteria for what constitutes sci-
ence yet put on the public record. The Overton criteria are:

• It [science] is guided by natural law.

• It has to be explanatory by reference to natural law.

A WARM LITTLE POND

125

• It is testable against the empirical world.

• Its conclusions are tentative, i.e., are not necessarily the final

word.

• It is falsifiable.

Needless to say, creation “science” fails to meet even one of
these criteria; ergo, as a scientific explanation for the origin of
life, it has no real place in our deliberations here.

If creation science has no role in a scientific consideration of
life’s origins, why have I devoted any space to it at all? Princi-
pally because the creationist controversy illustrates in the stark-
est possible terms the psychology and tactics of pseudoscience as
considered in abstract terms in Chapter One. All of the hall-
marks of pseudoscience enumerated there show up in glori-
ous detail in the Arkansas case: appeals to myths, a casual ap-
proach to evidence, irrefutable hypotheses, refusal to revise,
and all the other by-now-familiar calling cards of the pseudo-
scientist.

From an intellectual point of view, perhaps the most interest-
ing aspect of creationism is not the “what” of its beliefs about
the way life got started, but the “why.” That is to say, why is it
that adherence to a literal reading of the book of Genesis holds
such great appeal for so many people? There must be something
more to it than just the odd beliefs of a fringe group of back-
woods hicks, as even well-educated, obviously intelligent people
like Henry Morris and Duane Gish are not immune to its attrac-
tions. In grappling with this puzzling matter, I can only con-
clude that the reasons, whatever they are, lie much deeper than
in the mere surface phenomena of a religious belief about the
origin of life on Earth. To my eye, creation science is only a
symptom of a far more fundamental disenchantment with sci-
ence in general, and the overwhelmingly dominant role it plays
in daily life. Many people obviously feel threatened by what they
see as the control that science has gained over their lives, and
many others feel mistrustful of the claims made by the pro-
science lobby about the improvement in their lives that science
will provide. And who can blame them, with disasters like Cher-
nobyl, the Challenger, Bhopal, and Love Canal serving as con-
stant reminders of science and technology run amok? So my
feeling is that the simple, straightforward belief in the word of

126

PARADIGMS LOST

God as written in Genesis serves as a comforting counterweight
for those of a certain fundamentalist persuasion. And as long as
people remain ignorant of the limitations of science and the fact
that science is carried out by ordinary human beings with all
their foibles and weaknesses, the creationists, like the rich, will
always be with us.

Through a poetic twist of cosmic fate, I happen to be writing
these words on Christmas Eve (1987), the main day of the year
for creationists to recharge their batteries in preparation for
another 365 days of jousting with those who take their Bible
reading a little less literally (whoops, 366 days — I almost for-
got that 1988 is a leap year). As a Christmas bonus to my scien-
tific readers, let me close this section by recounting what is
surely one of the more amusing sideshows associated with the
Arkansas trial.

In mounting their defense, the biggest scientific gun that the
creationists seemed able to muster was none other than Fred
Hoyle’s comrade-in-arms, Chandra Wickramasinghe. The crea-
tionists presumably requested his appearance because he had
suggested the intervention of a Creator to explain life on Earth,
although the silicon chip Creator he and Hoyle had conjured up
probably wasn’t exactly what Henry Morris and Duane Gish
had in mind. Anyway, after beginning his testimony with a few
well-chosen words about life being the product of a Creator,
Wickramasinghe began veering off the track, entering into a
long, meandering exposition of his views on comets, diseases,
and the rest of the extraterrestrial apparatus underlying the
H&W theories. He ended his testimony for the defense by stating
that he saw no way that a rational scientist could endorse the
notion of a global flood, or an age for the Earth of less than
1 million years. With expert witnesses like this, the creationists
surely didn’t need any enemies! In summarizing this farcical
testimony, Judge Overton remarked that he was “at a loss to
understand why Dr. Wickramasinghe was called in behalf of the
defendants.” On this sorry note ends not only the case for the
state of Arkansas, but also our own Defense case for the extra-
terrestrial origins of life on Earth. Sic transit gloria. Before
moving on to summary arguments, let’s call upon a few expert
witnesses on the functional activities of life in order to get a
little better perspective on the prospects for scientifically wrap-
ping up the foregoing claims and arguments.

A WARM LITTLE POND

127

THE LOGIC OF LIFE

In recent years the standard Primordial Soup Theory has come
under increasing attack, both as a result of newly acquired ex-
perimental data and as a result of some serious reexamination of
the methods and arguments employed by its proponents. Before
pondering a verdict on the competing cases, let’s pause for a look
at some of the points about the soup that give skeptics cause for
concern.

• Reducing atmosphere: There has been a mounting body of evi-
dence to support the claim that the early atmosphere was not
nearly as reducing as claimed (and needed) by the soup theo-
rists. Data has been presented showing the presence of oxygen-
producing life forms and oxidizing mineral species in rocks
more than 3.5 billion years old, as well as calculations showing
that a significant amount of free oxygen could have been pro-
duced by photodissociation of water. While not conclusive,
these results certainly cast doubt upon origins theories that
hinge critically upon a lack of free oxygen in the primitive
atmosphere.

• Macromolecule polymerization: Soup theories rely upon the as-
sembly of proteins and nucleic acids by a linking-up of many
individual amino acids or nucleotides. The kinds of such
“polymerizations” needed for life are subject to many compet-
ing reactions, with processes of destruction just as prevalent
as those of construction. Thus, any viable Soup Theory would
have to offer an explanation of how the constructive processes
dominated those tending to tear apart potentially useful poly-
mer chains.

• Monomer concentrations: The amino acids produced in Miller-
type experiments usually appear in very low concentrations;
ditto for the nucleotide bases formed in the kinds of experi-
ments done by Eigen and Orgel. These concentrations are far
too low to have had a plausible chance of leading to any mean-
ingful spontaneous polymerization.

• Investigator interference: A common occurrence in the origins
business is for an investigator to postulate some sequence of
chemical reactions needed to lead to life. He then sets up ex-

128

PARADIGMS LOST

periments that could plausibly lead to the production of the
necessary intermediate chemical compounds. As each com-
pound is produced, no matter how small the quantity, the ex-
perimenter then proceeds to the next step, assuming that the
needed elements from the earlier steps are available at what-
ever level of purity and in whatever quantity desired. In fact,
almost all laboratory prebiotic simulations involve illegitimate
interference of this sort by the experimenter, in which he or
she adjusts the experimental conditions so as to violate plausi-
ble hypotheses about what it was like on the early Earth.

Every item on the above list is a potentially fatal flaw in any
conventional Soup Theory. Let’s think more positively and sup-
pose we were trying to construct an alternative theory instead of
just looking for holes in existing proposals. What kinds of prob-
lems would we want our “Stew Theory” to address effectively?
Here are just a few:

• A possibly oxidizing primordial atmosphere

• A dominance of destructive over synthetic processes in the
prebiotic environment

• A short time interval of, say, 170 million years for the appear-
ance of the first life forms

• The presence of Precambrian rock deposits exhibiting no geo-
logical or geochemical evidence for any sort of hydrocarbon-
rich primordial soup

• Creation of a controllable and readily recognizable barrier be-
tween what laboratory experiments do when left to themselves,
and what they do when there is active interference by the ex-
perimenter

Creation of a theory satisfying the above list of desiderata is a
tall order indeed. And what would we have even if we did pro-
duce such a theory? Would any theory really tell it true as to
how life did originate here on Earth, as opposed to how it might
have originated? Are all such theories only Just So stories, or do
we really have, at least in principle, a fighting chance to unravel
the actual sequence of events that took place over 4 billion years
ago? These questions lead us into deep waters of philosophy, in
particular into consideration of the distinction between operation
science and origin science.

* * *

A WARM LITTLE POND

129

The usual view of theories of knowledge requires that a scien-
tific theory be able to (1) explain observed phenomena, (2) pre-
dict phenomena that have not yet been observed, (3) be testable
by further experimentation, and (4) be modifiable as needed by
the results of new experiments. In order to have even a ghost of
a chance of satisfying these criteria, the scientific theory must
set out to explain a recurrent set of events. The final condition
for a scientific theory is basically what separates science from
some of the baser forms of pseudoscience, and is a condition that
cannot possibly be met if experiments cannot be performed. Un-
fortunately for origins theorists, the one thing that everyone
seems to agree upon is that the origin of life on Earth was a
one-time-only event, hence outside the bounds of what is nor-
mally thought of as a scientific theory.

The foregoing split between a once-and-once-only event and a
set of recurrent phenomena is what separates operation science
from origin science. Operation science deals with explanations of
recurring processes like the passage of the Earth around the
Sun, the union of hydrogen and oxygen to form water, and the
flow of electrons through a resistor. In short, with natural pro-
cesses that are, in principle, repeatable. Origin science, on the
other hand, addresses the unique events in life: formation of the
universe, World War II, the painting of the Mona Lisa — and
the origin of life itself. Such events are not explainable by tradi-
tional scientific theories for the simple reason that they are not
subject to experimentation; thus they are not falsifiable and
hence not scientific. Or are they? Is there a loophole of some
kind that would somehow enable us to make a unique event re-
peatable, at least to the extent that sufficiently detailed experi-
ments can be performed about the event so that in a scientific
sense the event becomes repeatable? Prebiotic experiments of the
Miller type are crude, slow, plodding steps in this direction;
some think that the modern computer offers a possibility for far
more rapid progress.

Earlier we discussed the idea of using a WEES apparatus to
simulate the entire environment of the early Earth. The idea is
to create a miniature version of the primordial seas, atmosphere,
energy sources, tidal pools, and all the rest, then turn the system
on and see what happens. The problem with such a simulation is
the time factor: It’s estimated that the first living forms on
Earth took on the order of at least 170 million years to arise.

130

PARADIGMS LOST

What tenure-hungry assistant professor can afford to wait mil-
lions of years for something publishable to start swimming
around in such a device (even if the environmental parameters
are right)? The digital computer offers us two distinct pathways
to get around this time barrier — Material Mode and Formal
Mode.

In Aristotle’s epistemology, there are four causes for the ap-
pearance of worldly events, causes that Aristotle offers as expla-
nations for why things are as they are. The four complementary,
mutually exclusive, and collectively exhaustive causes are mate-
rial, formal, efficient, and final causation, of which the first two
are the most relevant for us at the moment. According to Aris-
totle, material cause explains an event’s or object’s taking the
form it does as a result of the material elements out of which it
is composed. On the other hand, the event or object also has a
plan according to which it is constructed, and this plan is com-
pletely independent of the matter out of which the object is
built. In Aristotle’s scheme of things, the plan constitutes for-
mal cause.

In all the origins work surveyed above, the focus has been al-
most totally upon what in Aristotelian terms would be consid-
ered material cause. All of the primordial soup theorists begin
by postulating some kind of material elements composing the
soup, together with a sequence of physical processes that plausi-
bly lead from the primitive material elements of the soup to the
first life form. By and large, the major points of disagreement
revolve about matters of material causation, e.g., the gases com-
posing the primordial atmosphere, whether the original life form
was silicon or carbon based, and so on. For questions of this
sort, the computer can be run in Material Mode and used in the
commonly accepted manner to simulate the chemical and physi-
cal processes postulated by particular models. We saw a good
illustration of Material Mode in the computational experiments
of Niessert, who used this mode to investigate the plausibility of
the random replicator scenarios of Eigen. When operating in
Material Mode, the computer’s main role is to act as a time accel-
erator, allowing the basic physical and chemical processes to un-
fold on a time scale thousands, if not millions, of times faster
than real time. Thus, with the types of supercomputers cur-
rently available, a Material Mode simulation of the WEES
might yield something interesting after, say, just a few years
instead of the hundreds of millions that might be needed in real

A WARM LITTLE POND

131

time. Important and interesting as Material Mode undoubtedly
is, to my mind the really interesting way to use computers to
study life is to use them in Formal Mode.

Looking at life through the spectacles of formal causation
means that we forget entirely about what kinds of matter living
objects are composed of, and instead turn our attention to the
functional, or logical, structure of living agents. In other words,
we focus upon those aspects of living forms that distinguish
them from nonliving objects, and ignore entirely the kind of
“stuff” that they’re made out of. Philosophers of biology gener-
ally agree that the functional activities distinguishing living
forms are three in number: metabolism, self-repair, and replica-
tion. Let’s look at how these activities can be formally repre-
sented by the logical interconnections linking them, independent
of material considerations.

On September 20, 1948, John von Neumann delivered a lec-
ture at Caltech titled “On the General and Logical Theory of
Automata,” in which he laid the foundations for a functional
theory of life. Yon Neumann’s interest at the time was in ex-
plicating the logical principles permitting construction of a ma-
chine that would be capable of manufacturing copies of itself if
placed in an environment sufficiently rich in the necessary raw
materials. Thus, at first glance it would appear that von Neu-
mann’s attention was focused on material cause. But first im-
pressions can be deceiving, and a reading of the paper makes it
transparently clear that what von Neumann was driving at was
to give a mathematically complete account of the different kinds
of functional activities that such a self-reproducing object
would need to have in order to be able to function. The particu-
lar material composition of such a self-reproducing automaton was
of little interest to von Neumann, and I’m sure he couldn’t have
cared less if one tried to build such a machine from aluminum,
glass, steel, or, for that matter, lox and cream cheese. What in-
terested von Neumann was the different functions that would
have to be incorporated and coordinated to achieve the proper-
ties characterizing life; in short, the logic of life.

What von Neumann discovered was that any self-reproducing
object must contain four fundamental components:

A. A blueprint, providing the plan for construction of offspring

B . A factory, to carry out the construction

132

PARADIGMS LOST

C. A controller, to ensure that the factory follows the plan

D. A duplicating machine, to transmit a copy of the blueprint to
the offspring

In living cells these properties are physically manifested,
roughly speaking, in the DNA (the blueprint), the process of
translation (the factory), the specialized replicase enzymes (the
controller), and the process of replication (the duplicating ma-
chine). It’s worthy of note that von Neumann discovered these
abstract properties necessary for any living form more than five
years before the far more publicly celebrated work of Watson
and Crick, which dealt with the very special case of the kind of
life we now see on Earth. Such are the wages of the theoretician,
especially one who solves “only” the general case!

The work of von Neumann and his successors shows that
everything that’s functionally important about life can be
represented as logical patterns that are in principle implemen-
table in a multitude of material environments. The simplest and
most entertaining illustration of this point is the well-chron-
icled game of Life, an elementary board game invented by
the British mathematician J. H. Conway. The playing field
of Life can be imagined as a flat sheet of paper extending in-
finitely far in all directions, with the sheet ruled off into square
cells like a chessboard without colors. At any particular stage of
play, a given cell is either alive (ON) or dead (OFF), the live
cells being filled in with a dot, say, and the dead cells being left
blank. According to the rules laid down by Conway, whether a
particular cell is ON or OFF at the next stage of play depends
upon the current state of those cells that are its immediate
neighbors in what is termed the Moore neighborhood, depicted in
Figure 2.10. The rules are very simple: The cell is ON if exactly
three of its neighbors are ON; it is OFF if it has zero, one, or
more than three neighbors that are ON (death from isolation or
overcrowding); it retains its current state if exactly two of its
neighbors are ON. Conway said that he originally set up these
rules as a guess between balancing the birth of new cells in a
rich, cooperative environment of social support, and the death of
cells by overcrowding or isolation. Let’s look at a few rounds of
play-

Figure 2.11 shows the histories of three generations of Life
patterns, all of which initially begin with three ON cells. The

A WARM LITTLE POND

133

8! H® 1

.-i 0m

FIGURE 2.10 The Moore neighborhood of a Life cell

reader will note that the first three triplets all die out, while the
fourth forms a stable configuration called a Block, and the fifth,
termed a Blinker, oscillates indefinitely.

For our purposes, one of the most interesting patterns in Life
is the so-called Glider, which is a pattern that repeats itself after
four generations, but in the process moves one square down and
to the right. A picture of the Glider is shown in Figure 2.12. In
the early days of Life, Conway conjectured that there were no
Life patterns that could grow indefinitely (i.e., would never die
out), and offered a fifty-dollar reward for the first proof or
counterexample to his assertion. A group at MIT claimed the
prize by displaying “the Glider Gun,” shown in Figure 2.13.
This configuration is a spatially fixed oscillator that resumes its
original shape after thirty generations. Within this period, the
Gun shoots off a Glider that wanders across the playing field and
encounters the configuration in the upper-right comer called an
Eater, which is a fifteen-generation oscillator. The Eater swal-
lows up the Glider without undergoing any irreversible changes.
Since the Gun oscillates indefinitely, it can produce an infinite
number of Gliders, thereby showing that there are configura-
tions that “live forever.” This fact refutes Conway’s conjecture.
What does all this have to do with formal models of life, as op-

1st Generation 2nd Generation 3rd Generation

FIGURE 2.11 Life histones of some triplets

Generation 1 Generation 2 Generation 3 Generation 4 Generation 5

FIGURE 2.12 The Glider

A WARM LITTLE POND

135

FIGURE 2.13 The Glider Gun

posed to special facts about Life? As it turns out, it has almost
everything to do with it.

In von Neumann’s automata setup, just as in the behavior of
real-life cells, information in the DNA is used in two quite dis-
tinct ways: as instructions to be interpreted (as in gene transla-

136

PARADIGMS LOST

tion), and as instructions to be copied (as in DNA transcription).
Consequently, if a Life pattern could be displayed that would be
self-reproducing and use the information describing itself in
just these two ways, it would be difficult to argue that such a
pattern was not “alive,” in the sense that it would then display
all the features needed to constitute a living form. Conway
proved that such a Life configuration exists, although to display
it explicitly would require a playing field the size of a small city
(Venice, for instance).

Conway’s self -reproduction proof is based on the observation
that Glider Guns, as well as many other Life objects, can be pro-
duced in Glider collisions. He then shows that large constella-
tions of Glider Guns and Eaters can produce and manipulate
Gliders to force them to collide in just the right way to form a
copy of the original constellation. The proof begins not by con-
sidering reproduction per se, but by showing how the Life rule
allows one to construct a universal computer. Since the Life uni-
verse consists of an array of ON-OFF cells, what this amounts
to is showing that one can construct a Life pattern that acts like
a computer in the sense that we start with a pattern represent-
ing the computer and a pattern representing its programming.
The computer then calculates any desired result, which would
itself have to be expressed as a Life pattern. For numerical com-
putations, this might involve the Life computer’s emitting the
requisite number of figures or, perhaps, arranging the required
number of figures in some prespecified display area. Conway
showed that the circuitry of any possible computer can be trans-
lated into an appropriate Life pattern consisting only of Guns,
Gliders, Eaters, and Blocks.

The second part of the Conway proof is to show that any con-
ceivable Life pattern can be obtained by crashing together
streams of Gliders in just the right way. The crucial step in this
demonstration is to show how it’s possible to arrange to have
Gliders converge from four directions at once in order to repre-
sent the circuits of the computer properly. The details of the
ingenious solution to this problem are much too complicated to
enter into here, but they provide the last step needed to complete
Conway’s translation of von Neumann’s self-reproduction proof
into the language of Life. This pioneering result opened the door
to the use of the computer to study abstract life in Formal
Mode. But what might such a study look like?

A WARM LITTLE POND

137

* • •

To the general public, the Los Alamos National Laboratory is
usually thought of as nothing more than a bomb factory, inhab-
ited by a collection of Strangelovian characters with visions of
megatons dancing in their heads. While some sections of “the
Labs” undoubtedly comply with this distorted vision of reality,
many of the activities underway there are of a far more benign
character, including a major effort linking the power of the mod-
ern computer with some of the deepest problems in modern biol-
ogy. As an outgrowth of some of this work in theoretical and
computational biology, in the fall of 1987 the Labs were host to
the First International Conference on Artificial Life, which of-
fered computer models of processes from protein synthesis to
plant growth, all in the spirit of Conway’s demonstration of
computer life. The meeting organizer, Christopher Langton, con-
cisely summarized the credo of the artificial-life community by
stating that such studies seek “the ghost in the machine; an es-
sence arising out of matter but independent of it” — in other
words, formal causation!

While there’s little room here to detail the program of the
“artificial lifers,” the essential ingredients are already present
in an earlier paper by Langton himself, where he takes up the
issue of how to use cellular automata (such as Conway’s Life
universe) to study real life of the organic, soft, squishy type
biologists love and cherish.

Langton’s paper argues that the primary functional roles of
the proteins and nucleic acids are as follows:

• Catalysis: The special proteins associated with mediating chem-
ical reactions are the enzymes, which do their thing by the pro-
cess of catalysis, speeding up chemical reaction rates
dramatically, sometimes by a factor of 100 million or more.
Thus, for all practical purposes, enzymes determine which re-
actions occur and which do not. Among the most important
properties of enzymes is the ability to recognize particular
structures and to bring about changes in them. Hence, the en-
zymes are the active agents in the logic of life.

• Transport: Proteins are the main vehicles carrying molecules
and ions around in the cell.

• Structure: Most of the cellular components and body tissues
are formed out of proteins.

138

PARADIGMS LOST

• Regulation: The primary agents for regulating the production
and interactions of biomolecules in the cell are the proteins. In
this role, they act mainly as messengers to initiate changes in
enzyme activity or protein synthesis.

• Defense: Proteins constitute the main agents (the antibodies
and immunoglobulins) by which the body fights invasion by
foreign objects. These functions involve the recognition of for-
eign agents, and the production of various molecular com-
pounds to tie up or break down the foreign invader.

• Information: The nucleic acids DNA and RNA provide the
main information store in the cell. Various polymerase en-
zymes covering the DNA strands initiate the transcription of
DNA to RNA, while other polymerases act to trigger the tran-
scription of DNA in the process of replication.

With these functional roles in mind, Langton indicates how it
might be possible to associate a cellular automaton rule similar
to Conway’s rule for Life with each activity. In this way we
could formally represent each of the functional activities of a liv-
ing agent with a logical “machine.” Linking these individual
machines together would then create an object that could be said
to represent a living agent, albeit an artificial one. In the Lang-
ton paper this idea is actually carried out to create an artificial
ant colony, whose behavior under appropriate conditions is
strikingly similar to that displayed by real-life ants. Under
other conditions, Langton’s vants (virtual ants) display lifelike
behavior quite different from what you’d see in your backyard
terrestrial anthill, but perhaps just the kind of activity that
might be seen in an ant farm on some planet orbiting Tau Ceti.
Who knows? Anyway, the point is that the creation of life in a
machine rather than a test tube offers almost unlimited vistas
for experimentation with origin-of-life theories that would be
temporally or physically inaccessible by any other means.

Before leaving this topic, let me take a moment for the benefit
of those who think of “life” in a computer as being merely a
hacker’s metaphor having no connection with ordinary “wet”
life, other than as a kind of computer game. This kind of thinking
is becoming increasingly difficult to defend, the recent spate of
“computer viruses” being the most dramatic evidence. And these
are definitely not the same kind of bugs most of us are familiar
with from the computer programming jargon. Put simply, a com-

A WARM LITTLE POND

139

puter virus is a piece of software that mischievous, and some-
times malevolent, programmers deliberately place in a set of in-
structions, say onto a game diskette or in a program available on
public-access electronic bulletin boards. As soon as the program
is loaded into a computer, the virus buries itself somewhere deep
in the system, with instructions to come alive when some set of
conditions is satisfied. For example, one famous virus was in-
structed to monitor the computer’s clock and then “wake up”
when the date showed it to be the birthday of the Apple Computer
Corporation. Upon awakening, this benign virus temporarily
took over the computer’s operating system and printed a birth-
day greeting on the screen. Other, less benign, viruses have been
reported that wipe clean data files, cause hard-disk crashes, as
well as produce a variety of other nasty effects. The point is that
once these things get into a system like a multicomputer network,
they take on what has every appearance of being a life of their
own. They can grow by moving from system to system through
the communication links in the network, and they act just like
biological viruses by appropriating the machinery of the network
for their own purposes. With these mischievous creatures now all
too real, I think the day is definitely over when one can scoff at
the idea of computer life as having no real meaning. To some
computer manufacturers, data center managers, and users, these
viruses are all too real for comfort.

Having listened to the sideline kibitzers, let’s get back to the
business at hand — coming to some kind of closure on the convo-
luted and confusing circle of arguments, hopes, dreams, and
schemes for how life got started on Earth.

SUMMARY ARGUMENTS

The path we’ve followed in trying to get a handle on the various
theories proposed to explain the origin of life has been a long
and tortuous one, going from the very down-to-earth ideas of
Alexander Oparin to the acts of faith of the creationists. Let’s
first summarize the competing arguments for the Prosecution
and the Defense.

Just to be perfectly clear on the conflict we’re addressing, let’s
begin our summary by restating the bone of contention. The
Prosecution’s claim is that:

140

PARADIGMS LOST

Terrestrial life had its origin as a consequence of natural physical
and chemical processes occurring here on Earth.

The Defense’s claim is just the opposite:

Terrestrial life either was imported to Earth, or did not come
about as the result of natural physicochemical processes.

Telegraphically, the arguments are given in Tables 2.1 and 2.2.

LIFE ORIGINATED ON EARTH!

PROMOTER

ARGUMENT

Eigen, Orgel

random replicators, hypercycles

Gilbert, Cech

self -catalytic RNA

Oparin

coaeervates

Pox

proteinoids

Dyson, Shapiro, Margulis

double origin, parasites

Cairns-Smith

clay

TABLE 2.1 Summary arguments for the Prosecution

LIFE ORIGINATED ELSEWHERE!

PROMOTER

ARGUMENT

(“natural origins ”)

Crick

extraterrestrial seeding

Hoyle and Wickramasinghe I

interstellar clouds and comets

(“supernatural origins ”)

Hoyle and Wickramasinghe II

silicon-chip Creator, diseases

Morris, Gish

creationism

TABLE 2.2 Summary arguments for the Defense

BRINGING IN THE VERDICT

On the specific question to be settled: “Did life originate on
Earth or did it come from somewhere else?” my verdict comes
quick and easy: The Defense is guilty of murder of the facts in
the first degree! To my mind, even the wildest schemes of the
Prosecution are vastly more plausible than the pipedreams, fan-

A WARM LITTLE POND

141

tasies, and totally baseless speculations of the Defense. In fact,
if I were the defendants’ counsel I’d strongly advise following
the path blazed by that well-known exemplar of modern political
sagacity Spiro T. Agnew, urging them to enter a plea of nolo
contendere and throw themselves on the mercy of the court. With
the exception of H&W Version I, none of the Defense argu-
ments are even in principle scientific, and they would hardly be
worth a footnote in any serious account of the origin of life were
they not being advocated by scientists of some repute, and ad-
hered to by such large, seemingly uncritical, followings. The en-
tire Defense case seemed to be aptly summed up in a newspaper
account I read recently about a movement afoot in Iran to de-
clare all car dealers and real-estate agents legally guilty of the
greatest sin on the books in modern-day Iran: “corruption on
Earth.” Rereading the off-earthers’ claims for life’s origins, I
thought it would be a delightful touch of theological irony to
have the ayatollahs, of all people, expand their horizons and in-
clude the entire Defense contingent beneath their legal umbrella.
But when it comes to picking and choosing among the many con-
flicting arguments of the Prosecution, things start to get inter-
esting again.

Of the many Prosecution claims and scenarios, I must confess
to a sneaking bias in favor of the Clay Theory of Cairns-Smith.
My reasons? There are many, but perhaps the most appealing is
that in contrast to the competition, it hasn’t been strongly chal-
lenged by any serious scientific arguments, and especially ex-
periments, against it. Of course, one could argue (and many do)
that all this means is that the theory is new enough and offbeat
enough that no one has really looked very hard at it yet. Maybe
so. But to my ears at least, it has a ring of plausibility missing
from the songs being sung by any of the competition.

First of all, the Clay Theory is explicitly a Dual-Origin The-
ory, one that easily accommodates my prejudice for life’s origi-
nating with the proteins and then moving on to the nucleic acids.
Somehow it just doesn’t ring true that the nucleic acids, which
are really just the big, fat molecular slobs of the cell, should
arise before the proteins, which are the actual doers. Thus, any
theory that postulates proteins first has a built-in advantage in
my mind, and the Clay Theory certainly qualifies for these
bonus points. Second, the theory requires no special materials
and no special environment above and beyond what could be ex-
pected on the ancient Earth. Finally, I like the idea of starting

142

PARADIGMS LOST

with some kind of low-tech solution to the problem of how to get
life going, and then shifting over to today’s high-tech mode once
things are up and running. As an additional selling point, the
Clay Theory doesn’t rely upon the kind of highly unlikely link-
ing-up of many amino acids and/or nucleotides called for by the
other theories, linkages that have formed the basis for any num-
ber of “devastating” critiques of origins theories by information
theorists and others of that ilk. All in all, in my view the Cairns-
Smith scenario provides a good lesson in how you should wield
Ockham’s razor in science to slit the throats of your opponents:
Simply offer an argument leading to the same conclusions, but
with fewer and simpler hypotheses. This is the essence of good
theorizing as well as good model building, and to my mind
Cairns-Smith has just done a better job of it than any of the
others.

IT'S IN

THE GENES

CLAIM:

HUMAN BEHAVIOR PATTERNS ARE
DICTATED PRIMARILY BY THE GENES

NATURE/NURTURE: SENSE OR NONSENSE?

A few years ago, in one of the most fascinating and disturbing
experiments in the annals of behavioral psychology, Stanley Mil-
gram of Yale tested forty subjects from all walks of life for
their willingness to obey instructions given by a “leader” in a
situation in which the subjects might feel a personal abhorrence
for the actions they were called upon to perform. Specifically,
Milgram told each volunteer “teacher-subject” that the experi-
ment was in the noble cause of education, and was designed to
test whether or not punishing pupils for their mistakes would
have a positive effect on the pupils’ ability to learn.

144

PARADIGMS LOST

Milgram’s experimental setup involved placing the teacher
before a panel of thirty switches with labels ranging from “15
Yolts (Slight Shock)” to “450 Volts (Danger — Severe Shock)”
in steps of 15 volts each. The subject was told that whenever the
pupil gave the wrong answer to a question, a shock was to be
administered, beginning at the lowest level and increasing in se-
verity with each successive wrong answer. The supposed “pupil”
was in reality an actor hired by Milgram to simulate receiving
the shocks by emitting a spectrum of groans, screams, and
writhings, together with an assortment of statements and exple-
tives denouncing both the experiment and the experimenter. Mil-
gram told the subject to ignore the reactions of the pupil, and to
administer whatever level of shock was called for as per the rule
governing the experimental situation of the moment.

As the experiment unfolded, the pupil would deliberately give
the wrong answers to questions posed by the teacher, thereby
bringing on various electrical “punishments,” even up to the
danger level of 300 volts and beyond. Many of the subjects
balked at administering the higher levels of punishment, and
turned to Milgram with questioning looks and/or complaints
about continuing with the experiment. In these situations, Mil-
gram calmly explained that the teacher was to ignore the pupil’s
cries for mercy and carry on with the experiment. If the subject
was still reluctant to proceed, Milgram said that it was impor-
tant for the sake of the experiment that the procedure be fol-
lowed through to the end. His final argument was “You have no
other choice. You must go on.” What Milgram was out to dis-
cover was the number of subjects who would be willing to ad-
minister the highest levels of shock, even in the face of strong
personal and moral revulsion against the rules and conditions of
the experiment.

Prior to carrying out the experiment, Milgram explained his
idea to a group of thirty-nine psychiatrists and asked them to
predict the average percentage of people in an ordinary popula-
tion who would be willing to administer the highest shock level
of 450 volts. The overwhelming consensus was that virtually all
the subjects would refuse to obey the experimenter. The psychia-
trists felt that “most subjects would not go beyond 150 volts,”
and they expected that only 4 percent would go up to 300 volts.
Furthermore, they thought that only a pathological, sadistic, lu-
natic fringe of about 1 in 1,000 would give the highest shock of
450 volts.

IT'S IN THE GENES

145

What were the actual results? Well, over 60 percent of the sub-
jects continued to obey Milgram up to the 450-volt limit! In
repetitions of the experiment in other countries — South Africa,
Italy, West Germany, Australia — the percentage of obedient
teachers was even higher, reaching 85 percent in Munich. How
can we possibly account for this vast discrepancy between what
calm, rational, knowledgeable men predict in the comfort of
their study, and what pressured, flustered, but cooperative
“teachers” actually do in the laboratory of real life?

One’s first inclination might be to argue that there must be
some sort of built-in “animal aggression” instinct that was ac-
tivated by the experiment, and that Milgram’s subjects were
just following a genetic need to discharge this pent-up, primal
urge onto the pupil by administering the electrical shock. A
modern hard-core sociobiologist might even go so far as to claim
that this aggressive instinct evolved as an advantageous trait,
having been of survival value to our ancestors in their struggle
against the vicissitudes of life on the plains and in the caves,
ultimately finding its way into our genetic makeup as a remnant
of our ancient animal ways.

An alternative to this notion of genetic programming is to see
the subjects’ actions as a result of the social environment under
which the experiment was carried out. As Milgram himself
stated:

Most subjects in the experiment see their behavior in a larger con-
text that is benevolent and useful to society — the pursuit of scien-
tific truth. The psychological laboratory has a strong claim to
legitimacy and evokes trust and confidence in those who perform
there. An action such as shocking a victim, which in isolation ap-
pears evil, acquires a totally different meaning when placed in this
setting.

Thus, in this explanation the subject merges his unique person-
ality and personal moral code with that of larger institutional
structures, surrendering individual properties like loyalty, self-
sacrifice, and discipline to the service of malevolent systems of
authority.

Here we have two radically different explanations for why so
many subjects were willing to forgo their sense of personal
morality and responsibility for the sake of an institutional au-
thority figure: genetic determinism versus Marxian environmen-
talism. The problem for biologists, psychologists, sociologists,

146

PARADIGMS LOST

anthropologists, and other “-ologists” of this ilk is to sort out
which of these two polar explanations is more plausible. This, in
essence, is the problem of modern sociobiology — to discover the
degree to which hard- wired genetic programming dictates, or at
least strongly biases, the interactions of animals and humans
with their environment, i.e., their behavior. Put another way,
sociobiology is concerned with elucidating the biological basis of
all behavior.

At first sight it may seem slightly preposterous to argue that
any human behavior pattern is forced upon us by our genes
since, after all, we are free-thinking beings having the power to
decide our actions for ourselves. Comforting as this prejudice
may be, there are plenty of arguments against it. A trivial exam-
ple is our need for sleep. No one can question that sleeping is a
behavioral pattern common to all humans, and furthermore it
gives every appearance of being completely determined by our
physiological makeup; i.e., it is genetic, not learned. You might
argue that sleeping is not the type of behavior pattern we have
in mind when we speak of exercising our “free will,” and that
we’re more concerned with human social behavior: aggression to-
ward others, mating and bonding patterns, religious and ethical
codes — in short, all the kinds of behavior that anthropologists,
psychologists, and sociologists find interesting. But even here
the Nature-versus-nurture question is far from clear cut as, for
example, when we consider the problem of schizophrenia. It’s
hard to deny that the actions of a schizophrenic fall into the
category of “interesting” social behavior. Yet there is fairly
convincing medical evidence to indicate that this malady is at-
tributable to chemical imbalances in the brain, i.e., to a genetic
misprogramming. Thus, the task of the modern sociobiologist is
to examine the balance between social behavior that is primarily
dictated by the genes, like schizophrenia, and behavior that is
overwhelmingly determined by our social and/or cultural envi-
ronment, like that of Milgram’s obedient automatons.

Since the arguments of the sociobiologist are based upon the
idea of behavior patterns emerging as a result of biological evo-
lutionary pressures, they are couched in evolutionary terms in-
volving concepts such as genotypes, phenotypes, selection,
adaptation, and so forth. Consequently, to explore the plausibil-
ity of a genetic basis for behavior, our first order of business
must be to establish the basic vocabulary of the Darwinian evo-

IT'S IN THE GENES

147

lutionist, and then to look at how these biological notions fit to-
gether with the concepts of social behavior as seen by the etholo-
gist, sociologist, anthropologist, and psychologist. It is to this
that we now turn.

NEO-NEO-DARWINISM AND SOCIOBIOLOGY

The Central Dogma of Molecular Biology asserts, roughly
speaking, that there is a one-way flow of information from the
genes to an organism’s structural form. In short, we have the
chain DNA — RNA -* Proteins. For the purpose of studying
the implications of biology for behavior, we might profitably ex-
pand this pillar of molecular biology into what I’ll call the Cen-
tral Dogma of Social and Behavioral Biology, whose essence is
depicted in the following diagram:

Genotype

Environment

Form

=> Phenotype -» Function

•>*

Behavior

The Central Dogma of Social and Behavioral Biology

Since more than a minor amount of the rhetoric surrounding
the aspirations and claims of the sociobiologist arises from ter-
minological confusions involving the components of this dogma,
let me now pick apart the diagram and give a more detailed ac-
count of how each of its pieces is to be understood within the
context of our concerns in this chapter.

• Genotype: By far the most vexing terminological confusion in
the sociobiology literature surrounds the many and varied us-
ages of the term gene. In strict biochemical terms, the gene is
rather unambiguously defined as a section of the DNA strand
needed to code for the production of a single protein. How-
ever, when we pass beyond the borders of molecular biology
and begin moving toward “genetic” determination of behavior,
the concept becomes increasingly fuzzy. Since virtually all in-
teresting physical characteristics and behavioral traits involve
the cooperative action of several “genes,” as the term is used

148

PARADIGMS LOST

in its molecular biological sense, it has been suggested for soci-
obiological purposes that the word “gene” be replaced by the
term replicator, which is taken to mean the unit of genetic ma-
terial that we use when we refer to a Darwinian adaptation’s
being beneficial to the organism. In this sense, a replicator can
mean a combination of individual genes that generate some ob-
served behavioral and/or physiological property of an orga-
nism. With this idea in mind, we’ll consider an organism’s
genotype to be the totality of replicators contained in its
physicochemical genetic makeup.

• Environment: In our discussions, the term environment will al-
ways refer not only to an organism’s physical surroundings,
such as terrain, climate, water, and air, but also to the social
and cultural setting within which the organism carries on its
life activities. So, for instance, within this extended definition
of the everyday idea of what constitutes the environment, we
would say that the identical twins Jim and Joe had the same
genotype but different environments if Jim was a Hare
Krishna and Joe was a practicing Orthodox Jew, even if they
both lived in the same house and otherwise shared the same
life-style.

• Phenotype: Quite simply, an organism’s phenotype is the en-
semble of all of its observable physical, functional, and behav-
ioral characteristics, i.e., form, function, and behavior. Thus,
physical properties like color, size, and shape are part of the
phenotype, as are functional activities such as flying for birds
or swimming for fish. In addition, an organism’s phenotype
includes various behavioral traits characteristic of the orga-
nism, like hunting in packs for hyenas, pair bonding for pi-
geons, and the organizational patterns of social insects like
ants, bees, and wasps, not to mention cultural traits like paint-
ing or music for human beings.

With the foregoing ideas in mind, let’s now look at the pro-
cesses that compose today’s souped-up version of Darwin’s vi-
sion of evolution. In compact terms, we can express the
essential features of neo-Darwinian evolution by means of
Darwin’s Formula:

Variation + Heredity + Selection = Adaptation

As with our Central Dogma, each of the terms in Darwin’s
Formula requires amplification and elucidation.

IT'S IN THE GENES

149

• Variation: In the neo-Darwinian world, the term variation is
employed to refer only to change at the level of the organism’s
genotype. Such genotypic variations (which can be caused by
many environmental factors, such as temperature, radiation,
or just random mutations) may give rise to phenotypic differ-
ences.

• Heredity: In order for genotypic changes to be passed on to
offspring, it must be assumed that there is a mechanism by
which the parental genotypes are somehow transmitted to
their children. Since the idea of a gene was unknown in Dar-
win’s time, this problem of heredity was a major puzzle for
Darwin; nowadays we know that it is the replicators that are
passed on from one generation to the next by moving from one
temporary phenotypic host, or “survival machine,” to another.

• Selection: Not all phenotypes are created equal, and the crux of
the Darwinian scheme is the argument that Nature picks and
chooses among the phenotypes, bestowing on some the “right”
to produce more offspring than others. It’s crucial to note here
that although the phenotypic variation has its root cause in
changes in the genotype, the traditional Darwinian selection
mechanism acts only at the level of the phenotype. Further-
more, the decision “thumbs up/thumbs down” on a particular
phenotype is determined by the environment in which the phe-
notype is operating. Thus a thick coat of white hair has a
strong positive selective advantage for a polar bear at the
North Pole, but would work in just the opposite direction
should the same bear be transplanted to the Philippines.

• Adaptation: By definition, a phenotypic trait is termed adaptive
if possession of the trait gives an organism a reproductive ad-
vantage in its operating environment. Note again that a par-
ticular trait is never adaptive or maladaptive in and of itself;
its level of adaptation is always determined with regard to a
specific environment.

At this point in our deliberations, it’s useful to stop for a few
comments setting our terminological usage into perspective
within the mainstream socio biological literature.

First of all, the matter of fitness. I have avoided using this
term since in the literature it is often used more or less inter-
changeably in two quite distinct (and far from equivalent) ways.
The popular usage in Darwin’s time was what today we call phe-

150

PARADIGMS LOST

notypic fitness, which refers to the measure of an organism’s abil-
ity to survive and reproduce in a given environment. Note that
this criterion of fitness refers only to the organism’s phenotypic
characteristics, and says nothing about the genotype. Darwin
termed the process by which Nature rewards those of higher
phenotypic fitness natural selection. On the other hand, we have
the currently more fashionable idea of genetic fitness, which is a
measure of an organism’s genetic contribution to the next gener-
ation, i.e., how many copies of its genes find their way into the
gene pool of the next generation. This concept of fitness makes
no reference to the organism’s phenotypic properties at all.

With these two very different measures of fitness available, we
have to be very careful to make clear which one we’re using
when we begin waving our magic wand of evolution and start
talking about “selecting” an organism for reproductive advan-
tage. Of course, it could be claimed that the two measures of
fitness are highly correlated, using the argument that high phe-
notypic fitness gives an organism a leg up on the competition,
thereby enabling it to push more of its genes forward into the
next generation. On the surface this claim appears airtight, but
we’ll see later that it’s very difficult to explain how certain well-
established behavioral traits like altruism could ever arise if
such an argument were valid. The crux of the counterargument,
which we’ll also take up in detail later, is that such behavioral
traits could arise “naturally” only if we shift the focus of our
concept of fitness from the phenotype to the genotype. As we’ll
see, this shift in direction serves as a major plank in the plat-
form of most sociobiologists.

A second point to take note of, as indicated earlier, is that
Darwin knew nothing about genes or the precise mechanism by
which phenotypic fitness could be passed on to offspring. And, in
fact, such knowledge was not necessary for the arguments he
was making. All that was required was that there be some (not
necessarily perfect) correlation between the phenotypic proper-
ties of parents and offspring and the reproductive contributions
of each to future generations. In other words, Darwin needed
only a positive correlation between parents and offspring in
overall phenotypic fitness, without having to worry about the
precise mechanism by which this correlation came about.

Before moving on to a discussion of sociobiology per se, let’s
return to the Central Dogma for Social and Behavioral Biology

IT'S IN THE GENES

151

and carefully delineate just what it means to say that a behav-
ioral trait follows from a particular genotype. To begin with, I’d
like to dispel the simplistic, popular-science view that somehow
the cellular genetic material acts as a blueprint for assembling a
body from a set of individual pieces. While in molecular biology
it is true that a given gene corresponds to one and only one pro-
tein structure, there are a large number of poorly understood
steps between a bag full of proteins and a fully assembled, func-
tioning, living organism. Richard Dawkins has appealingly com-
pared DNA to a recipe for baking a cake from a set of raw
ingredients. With minor exceptions, there is no one-to-one corre-
spondence between the words of the recipe and the “bits” of the
cake. While the whole recipe maps onto the whole cake, if we
change one word of the recipe and bake one hundred cakes with
the original recipe and one hundred cakes with its “mutated”
version, what we will note is a consistent difference between the
two types of cakes, a difference that can be attributed to that
single change in the recipe. It is in exactly this sense that we can
say that genotype => phenotype in a fixed environment, and it
would be not only misleading but generally just plain wrong to
assert that there is any single “bit” of the organism’s genotype
that corresponds directly to any particular phenotypic charac-
teristic, including behavioral traits.

On this same general issue of genetic “determinism,” care
should be taken not to confuse the gene action involved in the
physical development of an individual organism from a fertilized
egg to a mature adult, a process that indeed does follow in a
causal manner from genotype to phenotype, with the kind of
acausal relationship between genotype and phenotype used in
population genetics. In the latter case, a proportion of the phe-
notypic variation observed in a population is “attributable” to a
correlated variation in the population genotype, with no claims
being made as to the causes of that correlation. For instance, we
might have a group of rats in which half have long tails, the
other half tails of normal length. Upon examining the genetic
makeup of the population, we may find that 60 percent of the
long-tailed rats have genotype X, while the rest of the popula-
tion are of genotype Y. In the population-genetic sense, we
would say that there is a positive correlation between genotype
X and the phenotypic property “long tail,” but we would not
necessarily infer that the presence of genotype X “caused” a

152

PARADIGMS LOST

long tail in any particular individual. In fact, we could not infer
this since 20 percent of the population display the alternate
genotype Y and yet still have long tails.

As we wend our way through the labyrinth of arguments of-
fered by the sociobiologists and their critics, the reader should
continually be on the lookout for the various ways in which the
above concepts and notions are employed. As noted, the litera-
ture is rampant with confusion on this score, and in many cases
the only way to make sense out of some of the verbal bombshells
flying about is to examine carefully the specific ways in which
the disputants are using these overworked everyday words and
ideas. With these caveats in hand, let’s now take a brief look at
the general framework of the research program of the sociobiolo-
gists before we move on to consider their ideas in all their elabo-
rate detail.

As a compact statement of the aims and claims of sociobiology,
we can hardly do better than quote directly from the work of
Charles Lumsden and Edward O. Wilson, two of the main play-
ers in the contemporary game of sociobiology. In their 1981 book
Genes , Mind, and Culture, they state:

THE CENTRAL TENET OF HUMAN SOCIOBIOLOGY

. . . social behaviors are shaped by natural selection. . . . Those
behaviors conferring the highest replacement rate in successive
generations are expected to prevail throughout local populations
and hence ultimately to influence the statistical distribution of
culture on a worldwide basis.

The Lumsden-Wilson thesis can be translated into the following
steps:

1. Some phenotypic characteristics that we currently possess
were adaptive traits at some time in the past.

2. The appearance of these adaptive traits was strongly in-
fluenced by our ancestors’ genotypes.

3. The genotypes that influenced the favorable traits have there-
fore been selected for.

4. The genotypes that influenced the maladaptive traits have
died out.

5. The reason why we display favorable phenotypes today is the

IT’S IN THE GENES

153

widespread presence of genotypes influencing adaptive pheno-
typic traits.

I •

(Since the Lumsden-Wilson thesis is so central to understand-
ing the sociobiology debate, let’s restate its premises in slightly
less formal language. The links in the sociobiological chain of
argument are strung together as follows:

Humans now display some kinds of behavior that were “good”
in the past.

These good behavioral traits are there because we inherited
them from our ancestors.

Therefore the good genotypes have been singled out for sur-
vival by natural selection.

The “bad” genotypes have been eliminated.

We have good behavioral traits now because the good genes
survived and the bad ones didn’t.

Providing the theoretical and experimental ammunition
needed to underwrite this chain of reasoning constitutes the
heart of the sociobiological research program. Needless to say,
the sine qua non of the program is the establishment of a tight
fit between the genotype and phenotype. A large part of our
story will be centered upon the nature of this fit and just how
tight it can be made.

To tie the concepts of phenotypic and genetic fitness into the
program of the sociobiologists, let’s call a behavioral trait
“phenotypically altruistic” if possession of that trait benefits
the survival of some other organism, while the trait is
“phenotypically selfish” if its possession benefits its owner’s own
personal survival. Similarly, we can say a behavioral trait is
“genetically selfish” if the effect of the behavior is to increase
the likelihood of the organism’s passing along copies of its own
genotype to future generations, while the trait is “genetically

154

PARADIGMS LOST

altruistic” if its effect is to increase the likelihood of genotypes
different from its own being passed on. With these distinctions
in mind, we can state:

THE STRATEGY OF SOCIOBIOLOGY

To explain all phenotypically altruistic behavior as being genetically

selfish acts

This section has introduced numerous terms and concepts that
will continually be referred to throughout the balance of the
chapter. So before letting the Prosecution loose with its argu-
ments supporting the research program of the sociobiologists,
let’s try to summarize the basic vocabulary in the following box.

TERMS AND CONCEPTS

replicator the unit of genetic selection influencing a pheno-
typic trait

genotype the totality of replicators forming an organism’s
biochemical genetic makeup

environment the physical, social, and cultural setting in
which an organism develops and lives

phenotype the totality of traits constituting an organism’s
form, function, and behavior

genetic fitness the relative ability of an organism to propa-
gate its genotype into future generations

phenotypic fitness the relative ability of an organism to sur-
vive in its current environment and reproduce

genetic selection the process by which Nature favors those or-
ganisms of high genetic fitness

phenotypic selection the “natural” Darwinian process by
which those organisms of high phenotypic fitness are fa-
vored by Nature

adaptation the process by which favorable traits (genetic or
phenotypic) are incorporated into the population

With the preliminaries out of the way, we now turn to the
advocates of sociobiology and ask them to present their case for
why we should believe that behavioral traits are governed princi-
pally by the genes. To avoid inflaming delicate sensibilities at the

IT'S IN THE GENES

155

outset, we will first consider the arguments for animals. Later
we’ll turn to a consideration of how relevant these results seem
to be for humans.

ANIMAL ANTICS

The literature surrounding the Darwinian Theory of Evolution
is filled with bizarre, crankish, and just plain incredible conten-
tions about the evolutionary pathway leading from apes to
humans. In this rogues’ gallery of craziness, surely the Yugo-
slavian Kiss Maerth takes the prize for batty ideas with his book
The Beginning Was the End: Man Came into Being Through Canni-
balism— Intelligence Can Be Eaten. According to Maerth, the
apes fed primarily on each other’s brains, and since brains are
an aphrodisiac, the apes’ gastronomic preferences increased
their sex drive, thereby whetting their appetite for more brains.
The most visible evolutionary result of this culinary “brain
drain” was the swelling of the apes’ own brains, making the apes
more intelligent. But Maerth claims that brain size increased at
a pace faster than the expansion rate of the skull, producing not
only migraines of gargantuan proportions for the apes, but also
an inflated view of their own importance in the overall scheme of
things. This, concludes Maerth, is why the state of mankind is in
its current deplorable mess. While it’s hard not to regard Ma-
erth’s evolutionary fantasy as a kind of scientific satire in the
style of Jonathan Swift, his line of reasoning does veer danger-
ously close to some of the arguments put forth by sociobiologists
wanting to infer by analogy human behavior from that of ani-
mals, especially the primates such as apes, monkeys, and ba-
boons.

One of the main taproots of modern sociobiology is the field of
ethology, or animal behavior, which was catapulted into promi-
nence when the 1973 Nobel Prize for physiology or medicine was
awarded jointly to Konrad Lorenz, Karl von Frisch, and Niko
Tinbergen for their well-chronicled studies of the imprinting of
geese, honeybee dances, seagull sex, and other types of animal
doings. Interestingly, it was the work of these men that formed
the starting point for a good bit of modern human sociobiology
(especially Lorenz’s studies of aggression). This is especially
ironic when we consider that both Lorenz and Frisch had been

156

PARADIGMS LOST

nominated much earlier for the prize, but turned down because
it was felt that their work did not apply directly to humans! As
we’ll see, there are many who still hold to this position today. Be
that as it may, this ethological work, coupled with the immensely
popular accounts of territoriality and aggression by Robert Ar-
drey, Desmond Morris, and Lorenz himself, set the stage for
today’s claims that there is something to be learned about
human behavior by observing the animals, and that that some-
thing involves social behavioral patterns placed into our geno-
type and passed on to us by our primitive animalistic forebears.
But just what sorts of animal behavior do sociobiologists have in
mind when pressing this extraordinary claim?

To the uninitiated, mention of the theory of evolution immedi-
ately brings forth the classic knee-jerk response “survival of the
fittest.” This catch phrase suggests, and rightly so, that an es-
sential feature of Darwin’s world is fierce competition between
species for limited resources of food, shelter, and sex. In short,
animal aggression — at least at the interspecies level. In the clas-
sic Lorenz-type studies on aggression, this kind of behavior is
correctly explained by appeal to natural selection, with the stud-
ies then going on to note that within a species there appears to
be only restrained fighting, usually involving ritual, bluff, and
violence of a nonfatal kind. According to Lorenz, these fights
within species are more like medieval jousting tournaments than
real wars, and are usually carried out for precisely the same rea-
son— winning the hands of the fairest maidens. For instance, in
ritual fighting between male bighorn sheep to determine which
will do most of the group’s mating, the contestants butt their
heads against each other until one of them signals his submis-
sion by baring his neck. At this stage the contest is over: The
victor retires to his newly won harem, while the loser limps off to
nurse his headache and, perhaps, to fight another day. Lorenz
claimed that aggression is instinctive; i.e., direct experience is
not necessary for it to develop normally. He also argued that
aggression is motivated by a “drive.”

To explain why there should be any fighting at all between
members of the same species, Lorenz offers the group selection
hypothesis: Such aggression exists to pick out the best (i.e., fit-
test) members of the group for breeding, since it’s in the group’s
overall interest to have its best members be parents. But it’s also
in the species’ best interest not to have any of its members
killed, since the weaker usually include the younger ones who

IT'S IN THE GENES

157

are needed to keep the species going in the future. This sort of
conventional-wisdom, group-selection-based scenario for animal
aggression has been challenged in almost every possible way by
the modern sociobiologist.

The sociobiologist’s first line of attack is at the level of the
facts: The almost universal principle of the limited nature of
aggression between members of the same species is far more fic-
tion than fact. Beginning at the level of insects and moving up
to the higher vertebrates, there is field evidence of case after
case of fights to the death, including even cannibalism, among
members of the same species. For example, lions sometimes kill
each other, and fathers are not beyond eating their cubs if given
the chance. Similarly, among chimpanzees, ants, and slugs we
see murder rates that make Las Yegas look positively benign.
And even birds display the sort of casual attitude toward mur-
der that most of us would associate with Colombian drug lords
rather than parakeets and blue jays.

At this point, one might begin to wonder how Lorenz could
have been so completely wrong. The sociobiologist has two an-
swers to this commonsense query: insufficient data and an erro-
neous theoretical foundation — the kind of one-two punch that
spells trouble for any purported scientific theory. According to
the guru of sociobiologists, E. 0. Wilson, it’s necessary to have
very long term studies of animal behavior to establish the full
truth about animal aggression, and Lorenz simply did not have
this kind of data. Wilson writes: “I have been impressed by how
often such behavior becomes apparent only when the observation
time devoted to a species passes the thousand hour mark.” He
then goes on to note that a murder every thousand hours is a
high level of violence by human standards, and that with the
more extensive data on animal behavior that is now becoming
available, humans are starting to look downright peaceful com-
pared with most of the animal kingdom, including the apes.

The second line of attack on Lorenz is directed against his the-
oretical group-selection hypothesis. The sociobiologist completely
rejects the concept of group selection, swearing allegiance only
to the notion that what’s good for the individual is ultimately
good for the group as well. Later we’ll try to provide solid argu-
ments for the adoption of this stance. For now let’s be content to
note only that individual selection is preferable to group selec-
tion, if for no other reason than by an appeal to Ockham’s razor:
It’s just simpler. With individual selection there’s no need for a

158

PARADIGMS LOST

priori assumptions about the good of the species, and as a conse-
quence there’s no reason to put forth special explanations for
why one member of a species would not attack another. Thus, all
things being equal, a lion is indifferent as to whether it’s attack-
ing a member of its own pride or a Thomson’s gazelle across the
savanna. After all, the first rule of survival is to survive. And
food is food, so you take it where you can find it.

Despite the limitations the sociobiologists put on Lorenz’s the-
ory, the sociobiological explanation of aggression still fails to ac-
count for the most surprising aspect of Lorenz’s studies:
Animals do show a remarkable degree of restraint in their con-
flicts with fellow members of the same species. The problem is to
offer an explanation based on individual selection for this ob-
served fact, as well as to explain why these conflicts sometimes
escalate. At one level such an explanation is trivial: Unre-
strained aggression must be more costly to the individual than
the exercise of restraint. But this is a pretty feeble sort of “ex-
planation.” At least this is what the eminent British biologist
John Maynard Smith thought when considering the question in
the early 1960s. He had the idea of looking at the problem of
animal conflict resolution as a “game,” employing ideas and
models originally pioneered by von Neumann and Oskar Mor-
genstern for the study of processes in economic bargaining.
Maynard Smith’s marriage of game theory to ethology has since
come to form one of the principal theoretical weapons in the
sociobiologist’s arsenal. Let’s see why.

The heart of Maynard Smith’s idea is the observation that in
any animal conflict, the respective payoffs to the individual con-
testants depend on the strategy employed by each of them. In
general there is no such thing as a uniformly best strategy, and
what a given individual should do in order to maximize his take
depends upon what his opponent is doing. Game theory enables
us to calculate what the optimal mix of actions would be in order
for a contestant to receive the greatest reward, on the average,
over a series of contests. To see the way things work, it’s best to
look at an example.

The simplest situation that illustrates the game-theoretic ideas
is the classic Hawk-Dove game introduced by Smith and Price
in 1973. The basic situation involves a population of animals
that are competing for some common resource. In any competi-
tion between two members of the population, each contestant has

IT'S IN THE GENES

159

the choice of opting for one of two “pure” courses of action:
Hawk, which is a policy of aggression in which the player always
escalates the battle until it is injured or its opponent gives way,
and Dove, a policy that begins with a traditional display and
then immediately gives way if the opponent begins to fight in
earnest. To make things as simple as possible, we further assume
that the members of the population reproduce asexually, and
that they breed “true,” i.e., offspring adopt exactly the same be-
havioral policy as the parent. Note that here we are implicitly
assuming a link between the genotype and the behavioral pheno-
type. We’ll come back to this crucial point later.

To measure the outcome of various interactions, let’s suppose
we have a unit of fitness V, which can be understood as the ex-
pected increase in an animal’s number of offspring if it can gain
the resource of contention without cost. Furthermore, when an
encounter escalates into a fight, the vanquished suffers a loss of
C units of fitness. Consider the possible types of conflict:

Hawk — * Hawk: In this case there is always a fight. The win-
ner gets all of the resource, while the loser is injured and disap-
pears. Since the situation is symmetric, any Hawk can expect to
win half its contests with other Hawks. Thus the expected
change in fitness for a Hawk is j( V — C).

Hawk «— 1 ’ Dove: In this case the Dove immediately runs away
at the first sign of Hawkish aggression, leaving the Hawk with
all the resource. In this situation the Hawk receives an increase
in fitness of the amount V, while the Dove gets 0.

Dove » Dove: In this peaceful situation of universal har-
mony and sharing, it can be expected that each “noncombatant”
will take the resource half the time, while giving it to the oppo-
nent the other half. In either case, the loser walks away unin-
jured and the expected gain in fitness to each is \ V.

We can summarize these expected payoffs for pairwise in-
teractions with the following array:

Hawk

Dove

Hawk

Dove

Here by convention the payoffs are to the player using the
course of action along the side against a player employing the
behavior along the top of the array.

Now imagine you are a member of the animal population and

160

PARADIGMS LOST

are faced with the decision to play Hawk or Dove. What should
you do if your goal is to maximize your overall payoff? Should
you always play one of the two pure strategies or should you
mix them in some proportion, sometimes playing Hawk and at
other times Dove? To address this question, we need the concept
of a strategy. Put simply, a strategy S is just a rule expressing
what fraction of the time a contestant plays Hawk and what
fraction it plays Dove. Thus, if the player adopts Hawk a frac-
tion p of the time and Dove a fraction q , then we can represent
this strategy as S = (p, q ), p + q =1.

At this point, Maynard Smith introduces a key idea enabling
us to calculate what the “best” choice of p and q would be. He
argues that the best choice would be those values of p and q that
lead to a strategy that is uninvadable. In other words, any ani-
mal playing a different strategy that tried to compete with one
playing this uninvadable strategy would, on the average, be
wiped out. Maynard Smith termed such a strategy an evolution-
ary stable strategy (JESS).

If the situation is such that the potential gain in fitness ex-
ceeds the cost of losing a contest, i.e., V > C, then it’s easy to
see that playing Hawk is an ESS, since those playing Dove
would meet mostly Hawks and would have a smaller payoff from
such encounters (0) than the expected amount of fitness increase
\ ( V — C) received by a Hawk encountering another Hawk. On
the other hand, playing pure Dove is not an ESS since Hawks
would have a field day in a population of Doves, gaining a dou-
ble payoff at every encounter, as opposed to the payoff they
would obtain in fighting another Hawk. But it’s probably more
realistic to assume that the cost of an injury is greater than the
benefits to be obtained from the contested resource, so let’s cal-
culate what the ESS strategy would be in this more interesting
situation when V < C.

To firmly fix these ideas, let’s plug in some numbers. Let p *
and q * be the values of p and q corresponding to an ESS when

V < C. For definiteness, suppose we have the situation in which

V = 5, C = 10; i.e., the increase in fitness acquired by winning
a fight is only half as great as the loss incurred by being de-
feated in battle. In this case it can be shown that p * — V/C =
^ = g. Therefore, the ESS is for a contestant to play Hawk
exactly half the time, Dove the other half.

There is an important technical point as to the interpretation

IT'S IN THE GENES

161

of the foregoing results that needs to be inserted here. We have
seen that no individual animal that plays a strategy different
from the ESS proportion between Hawk and Dove can survive
in the long run. Now suppose we have a population in which the
members cannot shift between Hawk and Dove at will, but are
constrained (genetically or otherwise) always to follow one of
the two courses of action. Question: Can we reinterpret the
above argument as saying that in such a situation it is evolu-
tionarily stable if a fraction V/C of the population plays Hawk,
while the remaining fraction 1 — V/C always plays Dove? An-
swer: Yes, if there are only two courses of action available to the
players; otherwise, the two interpretations lead to different re-
sults. This is just a mathematical oddity of the two-action situa-
tion, and has no deeper meaning in the context of the general
problem. Now let’s return to the question of the genetic basis
underlying these behavioral strategies.

Our earlier assumption of asexual reproduction ensured that,
given an equilibrium distribution of Hawks and Doves at which
the fitnesses were equal, the frequency of the offspring genera-
tion will be the same as the frequency of the parental generation
since the offspring are genetically identical to their parents. The
question is whether we can apply the same kind of game-theore-
tic arguments to sexually reproducing organisms like ourselves.
To address this question, consider the following example con-
structed by Philip Kitcher.

Assume we have an infinite, random-mating population of sex-
ually reproducing organisms with V = | and C = 1. In this
case, the ESS for the population is the strategy Indecisive, which
plays Hawk half the time, Dove the other half, just as in our
numerical example above. Suppose the initial state of the popu-
lation consists of individuals with three possible genotypes: A A ,
Aa, and aa, with AA animals playing Hawk, aa Dove, and Aa
Indecisive. Question: Is the strategy Indecisive of an Aa individ-
ual an ESS? Answer: No, as both pure strategies Hawk and
Dove can invade in the first generation and are maintained in
the population as a result of the sexual reproduction. In fact,
there is no ESS for individuals in this situation, although there
is a stable distribution of strategies for the population: | Hawk,
| Indecisive, j Dove. Thus, there is no way for an individual ani-
mal to move between the various actions and create an uninvada-

162

PARADIGMS LOST

ble strategy (an ESS), but there is a way for the population as a
whole to distribute itself so that no new population can invade.
This example should be kept in mind as we continue our discus-
sions later about the relevance of game-theoretic arguments for
social behavior. The moral for the moment is that the existence
of an ESS depends not only upon the available strategies and
payofls, but also upon the genotypes underlying those strategies.
Again it should be noted that this analysis assumes the existence
of such a genotype -* phenotype link.

The foregoing game-theoretic analysis has been pure armchair
speculation and back-of-the-envelope calculation. Does it have
anything to do with the way animals really behave in the wild?
Sociobiologists like David Barash have compiled considerable
field evidence that it does. One of the most interesting tests was
carried out by Susan Riechert, who studied the behavior of the
common grass spider A. aperta in settling territorial disputes.
Riechert studied these spiders in two habitats that differed
greatly in the availability of suitable locations for building
webs — a desert grassland in New Mexico and a desert riparian
area consisting of a woodland bordering a stream in Arizona, a
region offering many more favorable locations for webs. While
there is no room here to go into the details of how Riechert de-
termined the actions available to the spider and assigned the
various payoffs, her final conclusions are worth pondering. She
discovered that the contest behavior for web sites in the riparian
regions deviated substantially from the ESS predicted by the
game-theoretic model. In particular, contrary to theory, a
riparian spider does not withdraw from occupied territory when
it encounters the owner of the web. Rather, they engage in a
dispute that escalates to potentially injurious behavior. On the
other hand, the behavior of grassland spiders does follow the
ESS as predicted by the theory, with the time and energy they
expend in fights varying with their probability of emerging vic-
torious.

So while the riparian spiders are normally less aggressive
than their desert grassland cousins, just as ESS theory predicts,
they are still somewhat more aggressive than they should be.
This leads us to ask: Why does the behavior in these territorial
disputes differ from the ESS for riparian spiders and not for
their grassland cousins? Riechert gives an answer that will glad-
den the heart of any sociobiologist. She states:

IT'S IN THE GENES

163

If one assumes that the model is correct — that it has taken into
account all the important parameters and includes all possible sets
of strategies — then there must be some biological explanation for
the observed deviation. . . . One possibility is that the release from
strong competition is a recent event and that there just has not
been sufficient time for natural selection to operate on the behav-
ioral traits to complete the expected change. . . . Finally, a major
change in the wiring of A. aperta’s nervous system might be re-
quired to achieve the new ESS, and such a mutant may simply not
have arisen yet.

So far we have concentrated attention on animal conflict and
aggression as representative of the ideas and approach of the
sociobiologists to animal behavior. But at some stage the animals
have to stop fighting and start reproducing if their genes are to
be sent on to the next generation. In view of our earlier discus-
sion, let’s assume at the outset that this reproduction takes place
sexually, and take a moment or two to consider the process of
sexual selection and sex roles in animal mating from the sociobi-
ological point of view. A good case in point is the problem of
parental investment.

Both the male and the female want to produce children. But
production alone is not enough; someone has to bring up the
family. If one of the parents can off-load the work onto the
other, so much the better from an evolutionary standpoint, since
that parent is then free to go on the prowl for another mate with
whom it can produce more offspring. Naturally each parent
wants to adopt the same strategy, so the question arises of
whether the mother or the father has more to lose by adopting
the strategy of “hit and run.” Obviously, it’s normally the fe-
male that has more to lose if she decides to throw in the towel
and start over again. So there is a conflict of interest: The male
wants to “philander,” while the female wants not only to be fer-
tilized, but also to convince the male to hang around long enough
to help out with raising Junior. As a result we get different se-
lective forces at work, and what we expect (and usually find) is
that males tend to want to fertilize many females, while females
are more interested in raising those children that they already
have. To understand the sociobiological arguments underlying
these observations, let’s take a little closer look at the overall
situation.

164

PARADIGMS LOST

The key to understanding the evolution of the above kind of
sex role differences lies in the notion of parental investment. Ba-
sically, parental investment is any investment by the parent in an
individual child that increases the child’s chance of surviving at
the cost of the parent’s ability to invest in other offspring. Since
any parent has a limit on both the total amount of parental in-
vestment that it can make and on the total number of children
that it can have, we can work out the average investment per
child that an individual parent can make. By the definition of
sexual reproduction, each sex can produce only the same total
number of offspring as the other sex. But it’s not necessarily the
case that the two sexes in a species will have the same average
parental investment per child. As a result, the sex having the
greater average parental investment becomes a limiting resource
for the other sex. Figure 3.1 shows the situation graphically,
assuming that the female has the greater average parental in-
vestment. In this diagram, the female’s fitness is maximized
when she produces Of offspring, while the male’s fitness is high-
est when he produces Om offspring. Since Om is greater than Of
in this case males compete for females. Many of the territorial
disputes and aggressions discussed earlier arise for exactly this
reason: males seeking sexual access to females.

The story has been told so far from the viewpoint that selec-
tion acts only on the sex making the lesser parental investment.
But remember that the sociobiologist insists that selection acts
on the individual, so it must be the case that selective forces
are at work on the parent making the greater investment, too.
Just how could selection act to aid such a “giver”? The most
obvious way would be for selection to aid the giver by allowing
it to produce the largest number of the best possible children.
For the sake of discussion, let’s now assume that this giver is
the female.

In the terminology of Dawkins, there are at least two pure
strategies that such a giving individual could follow in looking
for a mate that would contribute to this Panglossian passel of
little savages: Domestic Bliss or He-man. The first involves the
female’s forcing the male to make a substantial investment
before copulation, a strategy probably all too familiar to sugar
daddys the world over. Under this strategy, the male is so com-
mitted by the time the children arrive that it might not pay him
to desert, because the next female he meets up with will probably

IT'S

IN THE GENES

165

NUMBER OF OFFSPRING PRODUCED
FIGURE 3.1. Parental investment and reproductive success

also demand such a priori efforts. Of course, this theory assumes
that the next female will indeed demand such efforts, so we must
be able to show that this behavioral trait will be an ESS strategy
in the population. Simple game-theoretic arguments very similar
to those of the Hawk-Dove variety show this actually to be the
case.

He-man is the other pure strategy open to the female. By
adopting this course of action, the female gives up on the idea of
having the male take out the garbage and bring home the bacon,
and settles for trying to get the best possible genes for her chil-
dren. Adoption of this strategy by the female places a high selec-
tive pressure on males to be strong, attractive, clever, and the
like, since this will be appealing to the female whose sons will
then be likely to carry these advantageous traits, thus giving
them a better chance of reproducing. Note that in the operation
of these female strategies, there will be a constant temptation
for males to appear fitter than they really are, with females try-

166

PARADIGMS LOST

ing to discriminate between those that are really fit and those
that just put on a good show. This observation led Wilson to
remark that under the He-man strategy, females would have a
strong tendency to develop coyness, i.e., hesitant and cautious
responses that evoke more displays from the male, thereby giv-
ing the female additional information with which to try to sepa-
rate out the “real men” from the “cads,” “flakes,” and poseurs.
Again, game-theoretic arguments can be used to examine the op-
timal mix between Domestic Bliss and He-man.

As the final stop on this whirlwind tour of the zoo, let’s look at
what for traditional Darwinists is one of the animal world’s
most difficult-to-fathom puzzles: the behavior of the sterile
worker castes in colonies of ants, bees, wasps, and termites. In
these settings there exist entire castes of sterile females who
devote their time exclusively to the well-being of their mother
(the queen) and their siblings. The British biologist William
Hamilton suggested the concept of kin selection in 1964 as a
mechanism to explain this otherwise highly non-Darwinian al-
truistic behavior.

Kin selection is based on the rock-solid premise that we are all
related to others. This means that each living creature shares
some of its genes with others, and since our genes have been se-
lected because of their ability to produce phenotypic character-
istics that assist their replication (or so say the sociobiologists,
anyway), it’s in our own selfish reproductive interest to see that
those to whom we are related reproduce. In short, only those
genes that reproduce persist, and the gene is indifferent as to
whether this is done directly or by proxy. Thus it might be
worthwhile to be altruistic to your otherwise useless, sponging
cousin because he will then be in a better position to pass on
some of your genes. As an aside, it should be noted that the idea
of kin selection goes back at least as far as another British biolo-
gist J.B.S. Haldane, who is reputed to have done a quick calcu-
lation on a beer mat in a London pub, coming to the conclusion
that he would gladly give up his life for three brothers or nine
first cousins. Here Haldane was simply following the rules of
Mendelian genetics, according to which he would share half his
genes with a full sibling, while sharing only one eighth of his
genes with a cousin.

The basic principle of kin selection can be generalized by the

IT'S IN THE GENES

167

rule: If the coefficient of relatedness (i.e., fraction of shared
genes) with another is r, and the benefit you can give to that
person in enhanced fitness for reproduction is k, then you
should give up your own chance at reproduction to help the
other if k > 1/r. So in the case of a full sibling (like Haldane’s
brother), r = \, implying that he should give up his own life to
save one brother if by doing so he could double his brother’s
chances of surviving to reproduce. Figure 3.2 shows how to com-
pute r for various degrees of relatedness. Each arrow in the dia-
gram means that there is a 50 percent chance that the two
individuals thus connected share genes. Hence, the likelihood
that any particular gene gets through n such arrows is (0.5)“.
When two individuals have more than one ancestor in common,
they can share genes via all of them, and we must then add all
possible paths. So, for example, for cousins we have

r=(aXbXcXf) + (dXeXcXf)

= (0.5 X 0.5 X 0.5 X 0.5) + (0.5 X 0.5 X 0.5 X 0.5)

= 0.0625 + 0.0625
= 0.125 (= |)

Hamilton’s contribution was to work out the mathematical de-
tails of the notion of inclusive fitness, which many feel is the most
significant extension of Darwin’s original idea since the incorpo-
ration of Mendelian genetics as the mechanism of heredity. Ac-
cording to Hamilton, the old Darwinian notion of individual
fitness (genetic or phenotypic) should be replaced by the individ-
ual’s inclusive fitness, which is defined as the individual’s own
personal fitness plus the individual’s influence on the fitness of
nondescendant relatives. There is no better way to see inclusive
fitness in action than to go back to the social insects and examine
Hamilton’s explanation for the appearance of the sterile worker
castes.

In the order Hymenoptera, which includes the ants, wasps,
and bees, the sex of offspring is determined in an unusual way.
Specifically, females are diploid, developing from fertilized eggs,
thus having both a mother and a father. On the other hand,
males develop from unfertilized eggs and are haploid, thus shar-
ing genes only with the mother (the queen). The result of this
odd sex-determination process is that sibling daughters of a
queen, fertilized by a single male, are more closely related to

168 PARADIGMS LOST

Aunt-niece

(or uncle-nephew, etc.)

FIGURE 3.2. Coefficients of relatedness

each other than they would be to any of their own daughters.
Graphically, the reason is depicted in Figure 3.3. Here the fe-
male Ego inherits two sets of genes: one from her mother, with
two sets, and one from her father, with one set. Hence the coef-
ficient of relatedness (average fraction of shared genes) between
Ego and a full sister isr = ^x| + |xl = |. But the coeffi-
cient between Ego and one of her daughters is only r = |. Thus,
Ego has more genes in common with one of her sisters than she
shares with one of her own daughters.

If Ego’s mother continues to produce cells for eggs after Ego
reaches maturity, then Ego will do the most toward perpetuat-

IT'S IN THE GENES

169

Offspring

FIGURE 3.3. Sex determination in the order Hymenoptera

ing her own genes if she devotes her time entirely to raising fer-
tile sisters, since fertile sisters will spread more of her genes
than will fertile daughters. More precisely, by adopting the cri-
terion of inclusive fitness, Ego’s self-interest is served if she
behaves “altruistically” toward her sisters rather than “self-

170

PARADIGMS LOST

ishly” for herself — in complete contradiction to what conven-
tional Darwinian genetic fitness would suggest.

Besides its intrinsic elegance, Hamilton’s explanation also sug-
gests why we don’t find any worker males: A male is no more
closely related to siblings than he is to daughters (he has no sons).
Another observation favoring Hamilton’s theory is that the nor-
mal 50-50 sex ratio found in animals that reproduce in the con-
ventional diploid manner, with genes contributed equally by
father and mother, is not seen in Hymenoptera. Hamilton’s the-
ory predicts that the ideal ratio of males to fertile females should
be one male for every three females, very close to what is actually
observed. Finally, there are the cases in which one colony takes
“prisoners of war” in a battle with another, with the queen then
able to make use of unrelated slave workers. In these situations,
the theory predicts a more normal 1:1 sex ratio, again exactly
what is seen in Nature. These results were taken to be convinc-
ing triumphs for the sociobiological arguments in favor of kin
selection and the notion of inclusive fitness. Needless to say,
however, they are not airtight and a number of difficulties have
been put forth casting at least a few shadows over the glowing
claims of the sociobiologists. We’ll look at these complaints when
the Defense takes the floor. For now, let’s try to summarize what
the sociobiological studies of animal behavior might suggest
about the relationship of genes, behaviors, and man.

As far as I can see, the basic chain of reasoning that human
sociobiologists would like to use from the study of animal behav-
ior consists of the following steps:

• In animals, especially those of the lower orders such as insects,
there is a close link between the genotype and phenotypic be-
havioral traits.

• Game-theoretic models based on the idea of maximizing inclu-
sive fitness give predictions in excellent accord with the way
animals actually behave in Nature.

• Extension of the classical notions of fitness by introducing
ideas of kin selection and inclusive fitness enables us to offer
good explanations for altruistic behavior in animals.

THEREFORE

• The same principles that work well to explain animal behavior
by genetic influence should work equally well to explain the
behavioral patterns of humans.

IT'S IN THE GENES

171

We will spend the rest of the chapter looking at the pros and
cons of this astoundingly ambitious chain of hopes and claims.

THE STRANGE CASE OF ALTRUISM

It’s been noted that a large number of winners of the Congres-
sional Medal of Honor have been soldiers who have thrown
themselves on hand grenades to save comrades. And in the ani-
mal world we have the honeybees, who buy themselves certain
death when they sting an intruder threatening the hive. How
can these acts of suicidal altruism be explained by the overtly
selfish principles of natural selection? This question has been de-
scribed as the central problem of sociobiology by no less an au-
thority than the head sociobiologist himself, Edward O. Wilson.
In the case of the social insects, we have already seen a fairly
convincing explanation of this altruistic behavior in the concepts
of kin selection and inclusive fitness put forward by William
Hamilton. But what about the many examples of human and
animal altruistic behavior that involve totally unrelated parties?
As a prelude to a full-scale examination of the arguments for
sociobiology, in this section we’ll devote our attention exclu-
sively to an investigation of the ways sociobiologists have de-
vised to say that “doing good for someone else can be doing good
for yourself.”

In the sociobiological literature, four distinct mechanisms
have been suggested to explain why an individual would take
actions decreasing his personal fitness in order to enhance the
fitness of another. We have already touched upon two of them —
group selection and kin selection — but for the sake of complete-
ness, let’s briefly review all four.

• Group selection: This was Lorenz’s explanation of why poten-
tially harmful aggression in animals appeared to be confined
to interspecies competition, and was rarely observed within
species. The basic idea is that an individual within a group
would be willing to suffer a personal loss in fitness if that loss
was more than compensated for by an increase in overall
group fitness. As a result of theoretical models, as well as inge-
nious alternative explanations, there is more or less universal
agreement today that group selection is a pretty rare phenom-
enon, taking place only under very special circumstances.

• Kin selection: We covered this explanation for altruistic be-

172

PARADIGMS LOST

havior between related individuals in some detail in the case of
the social insects, and the same notions seem to carry over
mutatis mutandis to humans. It’s often observed that close rela-
tives tend to look after each other more than they look after
strangers, and the closer the relationship (e.g., identical twins
versus distant cousins), the greater the willingness to sacrifice.

• Parental manipulation: This is a type of enforced altruism in
which a parent coerces a child to give help to another for the
parent’s benefit. A typical situation of this sort might arises,
for instance, if a mother cat has a litter of, say, five kittens
but can raise only three of them to maturity using her own
resources. Then it would pay her (genetically speaking) to em-
ploy her position of authority to force some of her older off-
spring to devote a part of their resources to helping her raise
the litter. She can do this in many ways, perhaps the most
common being a threat to withhold some of her attention from
certain offspring if they refuse to help out. In Nature the
strategy of parental manipulation often takes the form of can-
nibalism, in which the weaker members of the litter are sacri-
ficed for the benefit of the stronger. Of course it might be
argued that putting yourself on your brother’s dinner plate
hardly constitutes an “altruistic” act, in the sense that the
term is normally used in polite conversation. But in Nature
“altruism” means only an act that decreases your own fitness
in order to enhance the fitness of another, so such an act of
sacrifice is indeed altruistic, at least by Nature’s dictionary.

At first glance it may appear that there is no real difference
between parental manipulation and kin selection — they both
involve the sacrifice of an individual for the benefit of another.
However, there is one critical difference: In kin selection, one
individual helps another because they share some genes; in pa-
rental manipulation, one person helps another for the benefit
of a third party (the parent). The fact that the two parties
might share genes is incidental in parental manipulation, al-
though it often happens that they do. So in practice it may not
be easy to distinguish between the two forms of altruism, and
any given situation may involve both. In fact, it has been sug-
gested that the main causal factor at work in the development
of sterile castes in Hymenoptera is parental manipulation and
not kin selection. This is because when the queen sets up the
nest, she chooses to make workers rather than reproductives
by virtue of what she feeds her initial offspring. But this is

IT'S IN THE GENES

173

still a matter of some controversy and the jury is out as to
which of the two altruistic mechanisms is really at work here.

• Reciprocal altruism: By far the largest share of altruistic acts,
at least among humans, involve parties who are not related at
all. Robert Trivers introduced the idea of reciprocal altruism
to account for these sorts of sacrifical acts. In essence, the
principle governing reciprocal altruism is “If you’ll scratch
my back, I’ll scratch yours.” Briefly, the claim is that in-
dividuals engage in altruistic acts because they expect that by
doing so they will benefit by someone else’s altruism toward
them at sometime in the future. Note the very great difference
here between an act of reciprocal altruism and an act of kin
selection altruism. In the reciprocal case, the giver expects to
see a direct return from a sacrifice; in the latter situation, the
giver sees no direct reward but only the satisfaction of seeing
his or her genes being given a better chance to make it into
future generations.

The most convincing example of reciprocal altruism in Na-
ture seems to be the case of the “cleaner fish.” Certain species
of fish clean parasites off fish of a different species. This is a
situation in which both parties gain: The cleaners get a hearty
meal, while the fish being cleaned avoid the sores and diseases
that would otherwise result from the parasites. The most re-
markable aspect of this situation is that the cleaner fish are
never eaten by those they’re cleaning, even though this could
easily happen. Furthermore, it’s often the case that other types
of fish try to imitate the cleaners, rushing in to bite big chunks
off the fish being cleaned. In these cases, the big fish happily
gobble up the pretenders despite the fact that the pretenders
have developed high-level camouflage techniques to fool them.
Since the cleaners and the cleaned have no genetic relationship
at all, Trivers argues persuasively that this situation can be
explained only as a case of reciprocal altruism. We’ll return to
a deeper consideration of reciprocal altruism later on when we
consider the evolution of cooperative behavior.

THE GENETIC IMPERATIVE

From sad personal experience, I can attest to the fact that book
publishing, academic style, is a surefire prescription for ano-
nymity, totally unrewarding to anything but the ego. Only the

174

PARADIGMS LOST

fortunate few manage to sell even as many as a couple of thou-
sand copies of their magnum opus to libraries, captive students,
and a small band of fanatics and connoisseurs of the arcane. But
occasionally an academic author crashes through this paper cur-
tain of obscurity, putting forth a glitzy product with a market-
ing campaign rivaling that of the largest trade publishing
houses. Such was the case in the spring of 1975 when Harvard
University Press brought out Sociobiology: The New Synthesis, a
lavishly illustrated, seven-hundred page cofee table book by the
eminent insect expert Edward O. Wilson. In addition to full-
page ads in The New York Times Book Review, the book was the
subject of a front-page article in The New York Times describing
sociobiology as having “revolutionary” implications for human
societies. Similar statements were made in other major publica-
tions like People magazine, The National Observer, and The Bos-
ton Globe. What is it that gave Sociobiology and Wilson’s
subsequent book On Human Nature (which won a 1979 Pulitzer
Prize) their immense interest outside biology? Basically, it was
the extraordinary breadth of Wilson’s claims about the possibil-
ity of offering biological explanations for virtually all human so-
cial and cultural activities. Here we want to examine in some
detail both these claims and the arguments Wilson presents to
support them.

Wilson’s office at the Harvard Museum of Comparative Zool-
ogy is filled with colonies of various sorts of ants, the insects
whose behavior patterns started Wilson off on his path toward
trying to explain human behavior on the basis of biological prin-
ciples. As Wilson tells it, his books Sociobiology and On Human
Nature are really the second and third parts of an unplanned
trilogy that began with his 1971 classic The Insect Societies,
which, incidentally, was not on The New York Times best-seller
list! Wilson, a tall, thin Southerner in his late fifties, speaks
with great enthusiasm and verve about his passions (which in-
clude a firm commitment to jogging and a deep admiration for
people who have great goals and persevere toward them over a
long period). He talks about human sociobiology in just the
manner mentioned in the last section: as a natural extension of
the behavior patterns noted in animals. To understand his line
of argument as put forward in his books and subsequently re-
fined in numerous articles, interviews, and lectures, it’s useful to
think of the various steps in his program as rungs on a ladder

IT’S IN THE GENES

175

that must be climbed to reach his far-ranging conclusions. Our
version of this ladder paraphrases that originally put forward
by the philosopher of science Philip Kitcher.

WILSON'S LADDER

First Rung

Fitness maximization: Employing the usual methods of evolu-
tionary biology, we plausibly argue that all members of a popu-
lation P will maximize their fitness if they display behavior
pattern B in the typical environments faced by members of P.

Second Rung

Universality: If we observe that all members of P do in fact
display behavior B, then we can conclude that B became preva-
lent and remains so as a result of natural selection.

Third Rung

Selfish gene: If genetic fitness is used as the selection criterion,
selection can act only when there are genetic differences. Thus
we can conclude that there are such genetic differences between
the current members of P and their ancestors who did not dis-
play B.

Fourth Rung

Adaptation: Because there are genetic differences and because
B is adaptive, we can conclude that it will be difficult to modify
B by altering the social environment. This is because such an
alteration will be resisted by the i? -dominant population.

In Wilson’s scheme of things, we can identify three main lines
of attack supporting this ladder: gene inflation, analogy, and ad-
aptation. Let’s look at each in turn.

GENE INFLATION

This argument tries to assert the supremacy of the genes by
showing that the levels of biological organization that normally
mediate between the genotype and phenotype are either of no
consequence or are simply communication pathways for the ex-
pression of the genes. An eloquent advocate of gene inflation is
Richard Dawkins, whose book The Selfish Gene is a vastly enter-
taining, relentless pursuit of the idea that the organism is only

176

PARADIGMS LOST

DNA’s way of making more DNA. As an example of the kind of
logical tightrope that Dawkins walks, consider his distinction be-
tween the unit of selection and the process by which this unit is
singled out. He says: “If selection means differential survival
and reproduction, there is no question that it occurs between al-
leles [genes]. But the processes by which it occurs include differ-
ential survival and reproduction (selection) of individuals
[phenotypes].” Thus Dawkins asserts the supremacy of the
genes by assigning to the phenotype and the environment the
role of the mechanisms by which the genes are chosen. Oppo-
nents argue that it is misleading to imply inconsequential status
for the higher levels of biological organization, and that the
“selfish gene” argument fails to make a case for a tight geno-
type-phenotype fit because it tries to push out of the way the
most likely candidate for creating this gap in the first place: the
disproportionately large human brain.

ANALOGY

As has been noted, there are many human traits like sleep that
really are strongly determined by our genotype. Wilson’s argu-
ment by analogy states that if other behavioral traits are found
to be widespread across cultures, that fact constitutes a strong
prima facie case for there to be a substantial genetic component
underlying such traits.

As an example of this kind of reasoning about universal
human traits, Wilson offers the case of incest avoidance. Accord-
ing to Wilson, proscriptions against incest exist in virtually all
human cultures. His sociobiological explanation is that aversion
to mating with close relatives is a genetically programmed trait
that increases inclusive fitness, since inbreeding would have a
strong tendency to bring out lethal recessive genotypes. In fact,
Wilson goes further by citing the results of a study of 2,769
Israeli marriages in which none of the unions were between
members of the same kibbutz group raised together since birth.
Using this result, Wilson argued that the genetic tendency is not
just to avoid mating with blood relatives, but rather extends to
avoidance of sexual relations between members of any group
raised together since childhood. Wilson’s argument by analogy
is that the adaptive trait came about to prevent biologically unfit
offspring, and then “spilled over” to all close childhood associ-

IT'S IN THE GENES

177

ates. Skeptics ask, if the incest taboo is indeed universal and
genetic, why does incest need to be illegal?

ADAPTATION

Wilson writes as if he believes there are identifiable phenotypic
traits that are underwritten by specific “chunks” of genetic ma-
terial— what we have earlier termed replicators. He then goes on
to imply that any phenotypic trait that lasts must be adaptive,
and its adaptiveness must be explained by natural selection act-
ing so as to single out the underlying replicator. As an extreme
example, Wilson offers the religiously sanctioned cannibalism of
the Aztecs as a phenotypic response to the genetically pro-
grammed need for protein. Again a skeptic might say that such
a cultural response had nothing to do with genes for protein con-
sumption, but was due entirely to overpopulation of the environ-
ment. Another case of the same sort that Wilson puts forth
involves the widespread practice of homosexuality. How is it
that homosexuality could ever evolve as an evolutionarily advan-
tageous behavioral trait? Wilson’s answer is to appeal to the
notion of inclusive fitness, regarding the appearance of homosex-
uality as an adaptive response of the same sort as the appear-
ance of the sterile insect castes in Hymenoptera. That is, it
serves as a mechanism to prevent overpopulation. The general
problem with Wilson’s arguments from adaptation is that for
virtually every phenotypic trait, there are many Just So stories
that can be told for how that trait could have arisen as an adap-
tive behavioral response.

So we see that each of the main lines of argument Wilson puts
forth in his books Sociobiology and On Human Nature comes with
built-in, self-neutralizing counterarguments. Let’s briefly sum-
marize the main objections to his claims before taking a look at
how he tries to deal with them in later work. The principal flaws
in the early work appear to be:

• Underestimation of the power of the mind: Wilson continually
discounts the extraordinary power of the human brain to me-
diate between lower and higher levels of biological, social, and
cultural organization.

• Circularity: In Wilson’s claims, he assumes what he needs to
show, i.e., the causal path from the genotype to the behavioral
phenotype.

178

PARADIGMS LOST

• Isolatable traits: Wilson regards genotypic and phenotypic
traits as “atomic” units that can be isolated and studied indi-
vidually.

• Advantage versus adaptation: There is a continual confusion be-
tween traits that would be genetically advantageous, such as
banning weapons of war, and those that are the result of an
evolutionary adaptation.

In the five years or so following the appearance of Sociobiology ,
many of the objections noted above to Wilson’s line of argument
for human sociobiology came bubbling up out of the heated po-
litical, scientific, and philosophical debates surrounding his
sweeping claims. We’ll look at these debates in detail in a later
section, but it’s of more interest at the moment to see how Wil-
son, together with his colleague (and former student) Charles J.
Lumsden, tried to patch up the above gaps in their 1981 book
Genes, Mind, and Culture.

The main thrust of the Lumsden- Wilson position is aimed at
addressing the fundamental questions:

How much choosing do people actually do in the course of acquiring
or transmitting their cultural repertoire ? That is, how strong are
direct biases relative to other evolutionary forces acting on cultural
variation?

Where do the rules that direct choice come from and how do they
work?

In their book, Lumsden and Wilson try to give answers to these
deep matters by providing a mechanism through which the genes
can influence the development of mind, which in turn then acts
to produce culture. Finally, they close the loop by having cul-
ture act through natural selection to influence the genotype. The
claim is that this coevolutionary circuit closes the genotype-pheno-
type gap by way of the mind. Let’s examine the principal steps
in the Lumsden- Wilson circuit.

THE COEVOLUTION ARY CIRCUIT

1. Human culture consists of the interaction of all the ideas, in-
stitutions, behaviors, and artifacts used by a population.

2. We can use the term culturgen to mean an observable feature
of a culture.

IT'S IN THE GENES

179

3. During the process of forming a social order, the culturgens
are processed by epigenetic rules, which are genetically deter-
mined procedures that direct the formation of the mind.

4. The epigenetic rules of the mind bias the owner of that mind
to choose certain culturgens in preference to others.

5. The totality of all such choices in a population creates that
group’s culture and social organization.

6. Genetic variation takes place in the epigenetic rules, and this
variation accounts for at least some part of the variation in
behavioral choices that we see in a population.

7. Individuals whose choices increase their inclusive genetic fit-
ness are able to pass more of their genes along to the future
generations. As a result, the population as a whole is shifted
toward certain epigenetic rules and the types of behavior fa-
vored by those rules.

The entire Lumsden- Wilson circuit is schematically depicted
in Figure 3.4, showing the four main levels of biological organi-
zation. The molecular, cellular, and organismic steps constitute
the epigenesis, while the transition between the organismic and
populational levels involves the gene -* culture transition. The
final step of population influence on the genes takes place
through natural selection.

We can summarize the argument by saying that in this theory
the mind is formed out of a set of genetically determined rules
that bias it to choose certain interpretations of the world and
certain social and cultural options over others. Note the crucial
point here that what the genes prescribe is not a particular be-
havior, but only the capacity to develop certain behaviors and
the tendency to develop them in particular environments. In
other words, it is the epigenetic rules that are inherited because
the genotype actually codes for the construction of the wiring
pattern of the mind, which in turn encodes these rules. Thus, the
authors are claiming that the specific behavioral repertoire that
will be displayed depends on the experience that individuals re-
ceive within their own culture. So it is the total array of human
possibilities that is inherited, not the specific behavioral trait.

It’s fairly evident, I think, that all of the complaints leveled
against the early work of Wilson would vanish if the coevolu-
tionary theory could be established. Lumsden and Wilson state
the following conditions for such a validation of their theory:

180

PARADIGMS LOST

MOLECULAR

DNA

T-A-T-G-G-C-T

A-f-A-C-C-G-A

RNA

U-A-U-G-G-C-U

(Transcription)

a-u-A-c-c-g-a

PROTEIN

(Translation)

••• TYROSINE-GLYCINE-SERINE •••

POPULATIONAL

CULTURE

terms terms taste
terms

F I C U R E 3.4. The coevolutionary circuit

A. It must be shown that biased epigenetic rules exist.

B. It must be shown that these rules can be inherited.

C. It must be shown that we can establish a link between spe-
cific culturgens and inclusive genetic fitness.

D. It must be shown that there are molecular and cellular mech-
anisms that directly link the genotype to cognitive develop-
ment.

Surprisingly enough, there is evidence to support all four of
the above necessary conditions. To begin with, there do exist
biased epigenetic rules. For example, some people are born with
a clubfoot and would surely be biased against making the same

IT'S IN THE GENES

181

CELLULAR

ORGANISMIC

choice of footwear as those born with two normal feet. Further,
some epigenetic rules are clearly hereditable, such as the predis-
position to walk on two legs rather than on all fours. Thirdly,
some cultural choices do affect genetic fitness. For example, mak-
ing a living as a poisonous-snake handler or a movie stuntman
is likely to decrease one’s overall genetic fitness. Finally, there is
almost universal agreement that the code written in the DNA is
central to the construction and wiring of the central nervous
system.

So the Lumsden- Wilson Coevolutionary Theory is a con-
tender. The question really comes down to: How strong a con-
tender is it? How plausible is their argument compared with

182

PARADIGMS LOST

alternate interpretations of the same evidence? Just as a “for
instance,” the coevolutionary circuit operates only when the
phenotypic behavior modifies genetic fitness and when the pheno-
type is determined by the genotype. The problem is that there
are many candidate genotypes that could all lead to the same
phenotypic behavior. Other difficulties of this sort have also been
put forth against the blind acceptance of the coevolutionary the-
sis of gene determination of social patterns. As the eminent pa-
leontologist Stephen Jay Gould points out:

We have no evidence for biological change in brain size or struc-
ture since Homo sapiens appeared in the fossil record some fifty
thousand years ago. . . . All that we have done since then — the
greatest transformation in the shortest time that our planet has
experienced since its crust solidified nearly four billion years
ago — is the product of cultural evolution.

With the firing of the sociobiologists’ biggest gun, the coevolu-
tionary circuit, we complete the arguments supporting a biologi-
cal (i.e., evolutionary) basis for human behavior. Before letting
the Defense loose with its many-colored counterclaims, let’s first
get a feel for part of the Defense case by listening to just one of
the claimed excesses that Wilson has been accused of perpetrat-
ing— the support of sexism.

GETTING INTO HER GENES:

SEXISM AND SOCIOBIOLOGY

In a 1978 interview with Omni magazine, Wilson appeals to the
sex difference argument sketched earlier, claiming that there
currently exist “modest” genetic differences between men and
women that could be erased by careful training. As evidence he
cites studies of the second generation in an Israeli kibbutz,
where the regression of women to traditional roles was noted,
even in a social and cultural environment that explicitly called
for egalitarianism and equal opportunity. He then goes on to
state that there are three alternative courses of action open if we
want to tamper with this difference: (1) eliminate the difference;
(2) exaggerate the difference; (3) leave things as they are. His
claim is that by following the first course we could get statistical
equivalence of the sexes, but that it would require more knowl-
edge than we currently possess about the effects of gene manipu-

IT'S IN THE GENES

183

lation. On the other hand, Wilson argues that adoption of the
second course would only continue male domination and injus-
tice, stunting individual development. The third, laissez faire
course would most likely generate statistical imbalances in the
outcome, more or less like what we have today. He concludes
that there is probably no basis upon which a choice can be made,
and that there is a cost associated with each course of action.
Pretty reasonable, noncontroversial stuff, right? Wrong! It’s
statements like these that send Wilson’s critics into fits of apo-
plexy, running for their typewriters to denounce him for con-
tributing to the defeat of the ERA, as well as aiding and
abetting arch-conservative views that would deny the political
and social demands of those without power.

The heart of the argument that sociobiology is sexist is the
assertion that sexism is an outgrowth of the theory itself, at
least the version of sociobiology advocated by Wilson. The chain
of reasoning goes as follows: (1) socio biology begins by trying to
identify those traits that are common to people in all cultures;
(2) such universality is then taken to be an argument for the
trait’s genetic basis; (3) according to Wilson, one such trait is
an aggressive dominance system, with males reigning over
females; (4) therefore, sociobiology is inherently sexist. QED. In
fact, Wilson’s opponents have gone further and claimed that all
the important traits he identifies, like incest taboos, dominance
systems, and division of labor between sexually bonded pairs,
are based on sex differences. The problem, the critics argue, is
that Wilson is looking for a genetic cause, whereas what the
sociobiologist is really analyzing is adaptive function. But from
the standpoint of adaptive function, there is no difference be-
tween a behavior that is genetically programmed and one that is
culturally taught or individually learned.

Critics state that the underlying cause of sexism in sociobiol-
ogy is its basis in the kind of Darwinian sexual selection we de-
scribed earlier in our discussion of the Domestic Bliss versus
He-man strategies of mate selection. The argument against soci-
obiology is that this is only one of a number of possible forms of
natural selection, and its importance in the evolution of humans
is an untested hypothesis. One possible alternative, for example,
would be to claim that everything depends on the ecological set-
ting (environment), with an environment of abundance leading
to behavior that would minimize the social inferiority of the fe-

184

PARADIGMS LOST

male, while an environment of scarcity would generate behavior
emphasizing the sex role differences. A prime candidate for a
living example of this sort is found in the Tasaday of the Philip-
pines, a primitive tribe discovered in 1971 leading a Stone Age
existence with no concept of aggression or war, and also no ideas
about a male dominance hierarchy. Since all the Tasaday’s needs
were supplied by the lush Mindanao rain forest, the argument is
that this environment of plenty worked to create a social order
in which females and males participated equally.

A slight detour into sexism might be excused if it were seen as
an idiosyncrasy of an otherwise morally neutral study. But
when Wilson goes on to make claims for the sociobiological un-
derpinnings of such sensitive areas as homosexuality, religion,
ethics, and morals, members of the radical left, as well as a lot of
others, put on their gloves and come out swinging. We’ll hear
more from them in the next section. For now let’s take a moment
to let Wilson state his case concerning these delicate matters.

On the matter of religion, Wilson believes it is biological in
origin. He claims that religion is really the pivot of all that we
do and all that we really fight about, particularly when the reli-
gion becomes an ideology. In this view, religion is the one area of
behavior where you can’t draw any principles from the animal
world. Along with semantic language, it is the one truly human
trait and has to be considered as a biological property of hu-
mans, not just a cultural phenomenon or as the conduit for di-
vine guidance to man. Wilson’s hypothesis is that religion is
essentially an extension of tribalism and of our need to be able
to subordinate ourselves to concerted, irrational, even frenzied
group activity. The biological basis of this claim is a kin selec-
tion argument based on the principle that we all have genetic
predispositions toward xenophobia, attraction to charismatic
leaders, group worship, and so forth. Kin selection is then in-
voked as a mechanism whereby individuals subordinate them-
selves to the group for the overall welfare of the tribe.

The Wilson line on religion goes on to state that the religious
impulse is biological and uniquely human, but that religious
faith is almost always linked to imaginary scenarios and false
mythologies. He argues that it’s part of our biological predispo-
sition to make complete stories about the universe and the tribe,
stories that are always false and tend to be wiped out by science.

! T ' S IN THE GENES

185

Wilson concludes his overall argument by asserting that science
and religion will ultimately come together, with science adding
new depth to subjects that have traditionally been the province
of religion and the humanities. In this view, the new religion
will be a kind of scientific materialism, with competing ideolo-
gies like Marxism ultimately fading away.

From religion to morals and ethics is but a small step, one that
Wilson takes without a moment’s hesitation. His position is that
biological knowledge will help us arrive at a firmly based moral
code. His appeal to biology also includes the statement that bio-
logical principles will emphasize genetic diversity, at least till we
gain a much deeper understanding of human heredity. But then
Wilson undermines his own case by stating that even diversity
may not be a permanent value, leading his vitriolic critics to pro-
test that he is serving conservative, racist interests when he im-
plies that at some future time we may want to practice eugenics.

To understand Wilson’s position on these moral matters more
clearly, let’s examine what it is that he could mean by his gene-
based views on the relationship between sociobiology and morals.
According to Owen Flanagan, there appear to be at least four
different interpretations of Wilson’s vision:

1. Sociobiology can explain the origin of our moral capacities.

2. Sociobiology can explain the origin of particular moral beliefs
and practices.

3. Sociobiology can explain the basic nature and function of mo-
rality.

4. Sociobiology provides a way of generating certain normative
principles; i.e., it gives us a way of getting from “is” to
“ought.”

The first interpretation is trivially true; everything we do is
allowed by our genes, including the development of our moral
capacity. The second interpretation would imply that persisting
moral principles that enhance the genetic fitness of the group
that practices them must have genetic causes. This conclusion is
debatable, since it appears to be another instance of confusing a
trait that is advantageous with one that is adaptive. In this same
connection, Wilson argues that since morality evolved as a gen-
etic-fitness-enhancing trait, moral statements are statements
about genetic fitness strategies. To assess the merits of this ar-
gument, consider the following similar chain of reasoning:

186

PARADIGMS LOST

(a) our mathematical abilities evolved because they enhanced
our genetic fitness; therefore (b) mathematical statements are
statements about genetic fitness strategies. The most contentious
element on the list of possible interpretations of Wilson’s soci-
obiological view of morality is the last one. The claim is that we
are combinations of genes drawn from a pool, and that therefore
we should concern ourselves with the continued survival of
human genes in a common pool. Hence, we “ought” to act to
preserve the genes currently in the pool. But this argument ap-
pears to be little more than the statement that we ought to care
about the long-term consequences of our actions for the future.
If so, why do we need Wilson’s extra baggage of concern for the
genes? Why not just think about caring for persons and forget
the genes?

With the above questions on the table, we begin to edge away
from the arguments of sociobiology and into the territory of its
opponents. So without further ado, let’s give the floor to the first
Defense attorney, who will try to convince you not only that so-
ciobiology is pseudoscience, but that it’s positively politically
dangerous as well.

CANT VS. KANT

Shortly after the turn of the century, John D. Rockefeller, Sr.,
was busy pushing forward the interests of Standard Oil in a
manner that would make today’s antitrust lawyers salivate in
anticipation. Rockefeller was also a devout Baptist, and during
one of his weekly Sunday-school lectures he appealed to “natu-
ral law” as a means to justify his ruthless, predatory business
practices. On that occasion, he made the following oft-quoted
statement:

The growth of a large business is merely survival of the fittest.

. . . The American Beauty Rose can be produced in the splendor
and fragrance which bring cheer to its beholder only by sacrificing
the early buds which grow up around it. This is not an evil tend-
ency in business. It is merely the working out of a law of nature
and a law of God.

Good old J ohn D. — the very embodiment of the spirit that made
America great! Or so thought many of his contemporaries, a

IT'S IN THE GENES

187

number of whom also appealed to this Darwinian vision of the
natural order of things to salve their own consciences and, not
incidentally, to line their pockets.

In echoing their interpretation of Darwin’s universe, the John
D. Rockefellers of the world were following a path originally
blazed by the British philosopher Herbert Spencer, coiner of the
immortal phrase “survival of the fittest,” and a chief popula-
rizer of Darwin’s ideas. Ironically, Spencer himself was not a
Darwinist at all. He believed in the idea of Lamarckian inheri-
tance whereby phenotypic changes can directly influence the
genotype, in direct contradiction to the Central Dogma of Molec-
ular Biology. This is not, however, an unreasonable position to
hold in the context of social affairs, even if it is still anathema to
the molecular biologists. In today’s climate it’s hard to appreci-
ate the influence that Spencer’s social Darwinian ideas had on
the fabric of American life at the time, but a small indicator is
contained in a dissenting opinion given by the famed Supreme
Court justice Oliver Wendell Holmes, who stated that “the
Fourteenth Amendment [circumscribing governmental inter-
ference in the rights and actions of the individual] does not
enact Mr. Herbert Spencer’s Social Statics.” It’s just this sort
of social and political influence that the most vocal and rabid
of Wilson’s critics had in mind when they marshaled their forces
in 1975 to attack the claims they thought he’d made in Socio-
biology.

The Science for the People Sociobiology Study Group (the
Boston Group) is a collection of mostly radical-left scientists in
the Boston area whose most prominent members in the mid-
1970s were the eminent population geneticists Richard Lewontin
and Richard Levins, as well as the general public’s favorite pa-
leontologist, Stephen Jay Gould. Before carrying on, I should
emphasize that for the most part the members of this group
were internationally recognized scientists. Both Lewontin and
Levins were members of (or at least invited to join) the U.S.
National Academy of Sciences (although Lewontin resigned over
the matter of the academy’s issuing classified reports, while Le-
vins, a professed Marxist, refused membership because the acad-
emy engaged in military studies). Thus the attack that the
Boston Group mounted on Wilson’s scientific speculations is
particularly disturbing given the rarified academic reputations
of the group, and its spectacular disregard for the commonly

188

PARADIGMS LOST

accepted ground rules governing constructive (or even destruc-
tive) criticism within the ivy-covered walls and halls of aca-
demia. Following initially favorable reviews of Sociobiology, the
Boston Group issued a scathing attack not only on the book it-
self, but also on Wilson personally, linking him with the most
reactionary of political thinkers, including the Nazis. In addi-
tion, even though many members of the group were Wilson’s col-
leagues in the very same department at Harvard, the attack was
carried out publicly in a letter to The New York Review of Books
without even the courtesy of giving Wilson a copy of the criti-
cism prior to publication. Needless to say, this gross breach of
academic etiquette resulted in a spiraling escalation of attack
and counterattack that for a time even spilled over into the pop-
ular press. On the principle that where there’s so much smoke
there must be at least a few embers, let’s take some time to look
at the nature and content of these broadsides.

On the team of Philosophical All-Stars, Immanuel Kant is
definitely a heavy hitter. Unfortunately he wrote in a style that
confirms your worst fears about the writing of philosophers,
with even the most dedicated professor’s eyes glazing over when
slogging through one of Kant’s weighty tomes. But it’s heavy
going for heavy ideas, and with a little more luck and better
timing one of those grand-slam notions could have placed Kant
in the spotlight as the developer of sociobiology, rather than
Spencer or Wilson. One of the central tenets of Kant’s thought
is the categorical imperative, which, roughly speaking, is a claim
that humans have an innate awareness of moral law in the form
of a kind of rock-bottom ethical “ought.” By linking this Kan-
tian notion with our Central Dogma of Social and Behavioral
Biology, we come up with something that sounds remarkably
like Wilson’s sociobiological explanation for the evolution of
human ethics as an adaptive trait. Unfortunately for Wilson,
Kant never met Darwin, so when the Boston Group unleashed
its barrage of antisociobiological verbiage it was Wilson and not
Kant who was forced to take to the barricades. Doubly unfortu-
nate for Wilson was the fact that the group’s assault focused on
the cant of raw emotionalism rather than the Kant of pure rea-
son.

The essential content of the Boston Group’s letter to The New
York Review of Books was that Wilson’s book concealed a reac-

IT'S IN THE GENES

189

tionary political message. A direct quote expresses better than I
ever could the flavor of the political and personal attack:

These theories [biological determinism/sociobiology] provided an
important basis for the enactment of sterilization laws and re-
strictive immigration laws for the United States between 1910 and
1930 and also for the eugenics policies which led to the establish-
ment of gas chambers in Nazi Germany.

We think that this information has little relevance to human
behavior, and the supposedly objective, scientific approach in real-
ity conceals political assumptions. In his attempt to graft specula-
tion about human behavior onto a biological core Wilson uses a
number of strategies and sleights of hand which dispel any claim
for logical factual continuity. What Wilson illustrates to us is
. . . also the personal and social class prejudices of the researcher.

About a month later, Wilson replied to these charges:

I wish to protest the false statements and accusations that com-
prise the letter. . . . This letter ... is an openly partisan attack on
what the signers mistakenly conclude to be a political message in
the book. Every principal assertion made in the letter is either a
false statement or a distortion. On the most crucial points raised
by the signers, I have said the opposite of what was claimed. ... I
feel that the actions of Allen et al. [the group] represent the kind
of self-righteous vigilantism which not only produces falsehood
but also unjustly hurts individuals and through that kind of in-
timidation diminishes the spirit of free inquiry and discussion
crucial to the health of the intellectual community.

Let’s look a little deeper into the specific charges leveled at Wil-
son by the group, and the claim that his words and ideas had
been distorted.

The core of the Boston Group’s emotional outburst is orga-
nized around the assertion that Wilson is a biological determi-
nist whose work serves to buttress the institutions of society by
exonerating them from responsibility for social problems. In
support of these allegations, the group writes, “It is stated [by
Wilson] as a fact that genetical differences underlie variations
between cultures, when no evidence at all exists for this asser-
tion and there is some considerable evidence against it.” What
did Wilson really say? He wrote: “Even a small portion of this
[genetic] difference might predispose societies toward cultural
differences. At the very least, we should try to measure this

190

PARADIGMS LOST

amount. It is not valid to point to the absence of a behavioral
trait in one or a few societies as conclusive evidence that the
trait is environmentally induced and has no genetic disposition
in man. The very opposite could be true.” Not quite the same
either in content or in spirit as what’s claimed by the Boston
Group, is it? Another example? Well, try this one on for size.
The group writes that Wilson “promotes the analogy between
human and animal societies and leads one to believe that behav-
ior patterns in the two have the same basis.” In his book, Wilson
actually prefaces his discussion of this topic with the statement
“Roles in human societies are fundamentally different from the
castes of social insects.” This list of distortions and fabrications
could be considerably extended, but I think you get the general
drift. But how is it that the group could so consistently mis-
represent Wilson’s statements in the book? To answer this com-
monsense query, we have to dig a little deeper into the political
background of the group, especially that of its chief spokesman,
Richard Lewontin.

It’s no secret that Richard Lewontin advocates a Marxian
view of biology, and takes his professional job as a scientist as
tantamount to a political calling. He is on record with the state-
ment:

Any investigation into the genetic control of human behaviors is
bound to produce a pseudoscience that will inevitably be misused.
Nothing [emphasis added] we can know about the genetics of
human behavior can have any implications for human society.
But the process has social impact because the announcement that
research is being done is a political act. ... I treat my job as a
political activity.

Later he argued:

There is nothing in Marx, Lenin or Mao that is or that can be in
contradiction with the particular physical facts and processes of a
particular set of phenomena in the objective world.

One can only wonder why he omitted Stalin, Ho Chi Minh, and
Pol Pot from this list of infallible thinkers.

Faced with these outlandish statements by Lewontin, even the
most ardent zealot might start to cringe. But for us they offer a
window through which we can begin to see a bit more clearly
how it could be that the Boston Group would so blatantly twist

IT'S IN THE GENES

191

Wilson’s words and warp his meaning. To me it’s clear that the
group’s members were deeply alarmed by the impact that the
critical success of the book might have on the acceptability of
their own political views. Couple this fear with the inherent be-
lief that political philosophy should guide scientific research, an
attitude that was especially trendy on college campuses in the
1970s, and you have the basis for what Wilson once termed the
Fallacy of the Political Consequent. This fallacy consists of
the assumption that political belief systems can be mapped one-
to-one onto biological or psychological generalizations. Perhaps
not so surprisingly, this mapping points in exactly the opposite
direction from the one that Wilson’s tormentors accused him of
following!

In a later book written with Steven Rose and Leon Kamin,
Not in Our Genes, Lewontin continued his diatribe against Wil-
son by noting that sociobiology describes the whole of human-
kind as a transformation of European bourgeois society, with
Wilson’s description of human political economy involving a
possessive, individualist, entrepreneurial society that would cer-
tainly not apply to the serfs of Eastern Europe or Mayan and
Aztec peasants. Thus, in this view sociobiology treats categories
as if they were natural objects having a concrete reality, rather
than realizing that these are historically and ideologically condi-
tioned constructions. These authors then go on to level a per-
sonal attack upon Wilson, claiming that by emphasizing
altruism as a consequence of selection for reproductive selfish-
ness, Wilson has identified himself with American neoconserva-
tive libertarianism, which holds that society is best served if
each individual acts in a self-serving manner, limited only in the
case of doing extreme harm to others. In short, Wilson has
failed to separate out his “personal and class prejudices.” So in
this highly politicized view of reality, sociobiology is just the
most recent attempt to put natural science to work in the cause
of supporting the economic views arising out of Adam Smith’s
Invisible Hand, which, no doubt, in the opinion of Lewontin et
al. looks more like the Iron Fist. In fact, this whole business is
one of the most striking contemporary examples of the sociologi-
cal factor in science, which we discussed in the opening chapter.
Here we have strong cultural and political biases influencing not
only what is considered to be acceptable as a scientific research
topic, but also what is to count as scientific “truth.”

192

PARADIGMS LOST

As an illuminating final touch to this acrimonious exchange,
it’s of more than passing curiosity to note one or two of the per-
sonal factors surrounding the debate. Probably the most reveal-
ing background consideration of this sort is the close
relationship between Lewontin and Wilson prior to the dispute.
In fact, their offices at the Harvard Museum of Comparative Zo-
ology are only one floor apart. Moreover, it was Wilson who was
responsible for bringing Lewontin to Harvard in the first place,
over strong political opposition on the faculty. It was also Wil-
son who acted to promote Richard Levins ’s candidacy for the
membership in the National Academy of Sciences, which he later
turned down. So we see here a kind of family feud that unfortu-
nately bubbled over into the public arena dressed up as a scien-
tific debate. Despite the fact that these personal attacks forced
Wilson to cancel several engagements because of the mental
strain on his family, the group’s most publicly visible member,
Stephen Jay Gould, still had the audacity to remark that “we
don’t intend it as a personal attack. Ed Wilson is a colleague
whom we like.” Apparently the group members feel that it’s not
a personal attack to accuse someone of writing a book that is not
only totally valueless but even dangerous, because it’s filled with
what they claim are the author’s personal political views. Per-
haps the dictionaries in Cambridge offer a special definition of
what constitutes a “personal” as opposed to “professional” at-
tack, definitions differing from those found in the Worterbucher
I use in Vienna. In any case, Gould’s claim surely qualifies for
honorable mention at the next International Hairsplitters’ Con-
vention. Now let’s move on to more substantive scientific criti-
cisms of sociobiology.

SO-SO BIOLOGY

A few years after the publication of Darwin’s classic works, the
French writer l^mile Zola began his Rougon-Macquart cycle, a se-
ries of twenty novels described in a subtitle as The Natural and
Social History of a Family Under the Second Empire. This cycle
was intended to show the inevitable consequences of certain
scientific “facts,” especially the claims of Cesare Lombroso and
Paul Broca that inherited physical characteristics were indica-
tive of mental and moral traits. For example, in Nana, probably

IT'S IN THE GENES

193

the best-known novel in the cycle, Zola tells of the trials of the
courtesan Nana, the laundress Gervaise, and the drunk Cou-
peau. The complete family history is set up to dramatize Zola’s
statement that “heredity has its laws, just as does gravitation.”
It is this idea of biological (read genetic) determinism that con-
stitutes the main focus of the more sober, scientific criticisms of
sociobiology.

Technically, socio biology is based on a new view of natural
selection: Hamilton’s idea of inclusive genetic fitness. Implicit in
the claims of sociobiology is the notion that organisms act so as
to maximize their inclusive reproductive fitness. Critics have ar-
gued that this just is not true. Organisms act to maximize inclu-
sive fitness under constraints. The following list of such
constraints offered by Barry Schwartz indicates their impor-
tance in assessing the merits of the sociobiological case:

• Neutral characteristics: As far as genetic fitness goes, many
phenotypic properties of the organism are irrelevant, i.e., neu-
tral. Nevertheless, such characteristics may severely restrict
the kinds of future modifications of the organism that will
count as an improvement.

• Time lags: The processes of environmental change and evolu-
tionary adjustment operate on vastly different time scales.
Thus, what was optimal long ago may be very far from opti-
mal today.

• Context dependence: Genes leading to a certain kind of behavior
seen today may have originally come about for some quite dif-
ferent purpose, one that is no longer relevant in the current
environment.

• Historical constraints: Every modification must also be an im-
provement in order to avoid being eliminated. Thus, Nature is
totally oriented to the short term, performing local optimiza-
tions that may not lead to globally optimal performance. A
good illustration from technology of this kind of phenomenon
is the development of the transistor. No sequence of evolution-
ary improvements on the vacuum tube would have brought
this change about— a fundamentally new principle was needed.

• Variation constaint: Maximization can be applied only to those
variations that actually occur, not to those that were possible
but just didn’t happen.

• Cost-benefit analysis: Each of the subsystems composing an or-

194

PARADIGMS LOST

ganism has to coexist with every other, so that a variation
that’s good for one system may be very bad for another. Con-
sequently, every variation has to be measured by a cost-benefit
calculation against the overall improvement for the organism.
So, for example, the development of a capability to run faster
to catch food would have to be weighed against the extra en-
ergy needed to supply the added motive power.

• Levels of analysis: A given variation has to be evaluated at sev-
eral biological levels — gene, organism, group — and what’s good
for one may be very harmful to another.

• Capricious environment: A sudden environmental disturbance
can undo in a few days the gradual evolutionary changes of
several millennia, e.g., the meteorite collision that supposedly
wiped out the dinosaurs 65 million years ago.

The problem that this list of constraints poses for sociobiology
is that it’s difficult to give a criterion for what constitutes a
maximizing trait and what is only a background constraint. In
any particular situation, any activity at all can be shown to be
adaptive if the background constraints are drawn tightly
enough. Sociobiology adopts the strategy of arguing from be-
havior, and asks what the constraints must be so that a given
behavior maximizes inclusive genetic fitness. This is a circular
argument, and leads to Gould’s complaint that sociobiological
explanations are merely a collection of Just So stories.

The problem of genetic constraints is a special case of the
more general criticism that sociobiological explanations rely too
heavily upon the use of reification, i.e., treating idealized abstrac-
tions as if they were concrete entities. Specifically, the critics
contend that the sociobiologists systematically overrate the rela-
tionship between the genotype and various observed behavioral
traits. We have already considered some of these objections to
the assumed tight genotype-phenotype fit, and the attempts by
the sociobiologists to wriggle off the hook by introducing hypo-
thetical constructs like replicators or by arguing for genetic “in-
fluence” rather than determination. So let’s consider here a
different criticism, but one that pushes in the same general di-
rection.

It’s been seen that the explanation of altruism occupies a cen-
tral place in the theoretical framework of the sociobiologists,

IT'S IN THE GENES

195

and that the notions of kin selection and inclusive fitness are
crucial for the sociobiological argument to go through. The piv-
otal role in this argument is the determination of the coefficient
of relationship r expressing the genetic linkage between any two
family members. M. Sahlins contends that this is all mystical
nonsense, with the computation, or even recognition, of r being
impossible. The sociobiologist’s counterthrust is to concede read-
ily that the organism doesn’t sit down and explicitly calculate r
when deciding upon what action to take, but it acts as if it has
made such a calculation. And for sociobiological purposes that’s
all that counts. This is much the same argument that you might
use if someone claimed that you could never catch a baseball be-
cause you couldn’t solve the differential equations governing the
ball’s flight path. When set in this context, Sahlins’s assertion
starts to lose some of its initial luster.

It might appear surprising to some to invoke the name of
Richard Dawkins in connection with arguments against sociobi-
ology, but it has always seemed to me that his notion of a cul-
tural meme playing the role for cultural traits that genes play
for physiological ones is really a statement against the geneti-
cally based claims of the mainline sociobiologists. In the last
chapter of his book The Selfish Gene, Dawkins introduces the
meme, a kind of unit of selection for cultural matters of
the same sort that Lumsden and Wilson label a culturgen. The
memes are the carriers of such things as fashions, popular
tunes, and fads in speech, but unlike culturgens, memes are not
claimed to have a direct relationship with the actual genotype.
In fact, Dawkins goes further and argues that “memes and
genes may often reinforce each other, but they sometimes come
into opposition.” However, he does emphasize the functional
similarity of memes to genes in that both are replicators and
both are carriers of information. But when it comes down to
exactly how the information is carried and replicated, Dawkins
parts company with Wilson and Lumsden. Memes replicate by
being passed from brain to brain as pure information; cultur-
gens replicate by epigenetic rules processed by the physical
genotype. It is in this sense that I see Dawkins’s process of cul-
tural evolution as an antisociobiological argument against the
strict material transmission inherent in the hard-core position of
Lumsden and Wilson.

Now let me turn to one of the major objections put forward

196

PARADIGMS LOST

against treating sociobiology as a science. This is the old Pop-
perian criticism that in order to be scientific a theory must be
falsifiable, with sociobiology failing the test. The critic’s case
rests upon the Just So character of sociobiological claims that
allow virtually any behavioral pattern to be seen as an adaptive
trait. Thus, the argument goes, there is no conceivable observa-
tion or experimental result that would falsify sociobiology; ergo,
the theory is nonscientific.

There are at least two comments to be made regarding this
objection. First of all, it is just plain false to claim that there
are no observations that would falsify the theory. For example,
the observation of societies in which relatives gave freely with
no hope of return would surely deal a mortal blow to the aspira-
tions of the sociobiologists. The fact that no such societies have
been observed can hardly be laid at the doorstep of the sociobi-
ologists. Another example would arise if we were to see societies
that actively promoted incest. In this case, either (a ) the inci-
dence of birth defects is not significantly increased, or (b ) the
birth defect rate does rise, but the practice of incest goes on any-
way. The first alternative would refute the claim that incest
avoidance reduces inclusive fitness, while the second refutes the
claim that behavioral traits can be explained as enhancing gen-
etic fitness; i.e., it would be an example of a trait that persists
even though it has a negative effect on inclusive fitness. Either
alternative would spell deep trouble for the sociobiological view
of the world.

As a second point, we have already noted in the opening chap-
ter that there are serious difficulties with Popper’s falsification
criterion for separating science from pseudoscience. Interest-
ingly enough, a major stumbling block for Popper is the Prob-
lem of Auxiliary Hypotheses, a difficulty strikingly similar to
the Problem of Genetic Constraints noted above. Furthermore,
Kuhn has noted that much can be gained by not allowing a theory
to be blown away by the first fact that appears to contradict
it i-e > a strict adherence to the falsification ist doctrine can be
hazardous to the ultimate health of science! So with these ideas
in mind, the claims that sociobiology is not scientific also begin
to take on a distinctly less convincing air.

Finally, we move from falsification to the assertion that socio-
biology is just plain false. Here the critics contend that the gen-
etic differences between populations are not great enough to

IT'S IN THE GENES

197

account for the vast differences in culture that have been ob-
served. In this connection, the Boston Group states that “at
least 85 percent of that kind of [genetic] variation lies within
any local population or nation, with a maximum of about 8 per-
cent between nations and 7 percent between major races.” The
implication is that this relatively minor variation between na-
tions and races is way too small to be a significant factor in gen-
erating cultural differences.

The sociobiologists have two replies to this critique. First of
all, they argue on the basis of what Wilson calls the multiplier
effect, whereby small changes in the genotype can multiply and,
by the time they percolate up to the phenotype, can give rise to
major phenotypic variations. As might be imagined, the critics
look upon this kind of response with the same degree of favor
that small children look upon a plate of spinach. As always, the
Boston Group speaks with the sharpest tongue when it com-
ments that the so-called multiplier effect and its closely as-
sociated “threshold effect” are “pure inventions of convenience
without any evidence to support them. They have been created
out of whole cloth to seal off the last loophole through which the
theory might have been tested against the real world.”

By way of a second response, the sociobiologists employ a lit-
tle rhetorical judo, using their opponents’ strength against them
by turning their argument on its head. The sociobiologists say,
instead of looking at cultural differences, let’s look at cultural
similarities. The case is then made that the similarities are much
more important than the differences, and that these similarities
indicate a common genetic background. Of course, stating this
argument, just like stating the argument based on the multiplier
effect, is a far cry from providing a convincing demonstration
that it’s true, or even plausible.

On this inconclusive note we wrap up our quick survey of the
main epistemological objections to sociobiology. Since the central
pillar upon which the entire sociobiology program rests is the
explanation of how altruistic, or at least cooperative, behavior
can emerge out of basically selfish motives, it is worth spending
a moment looking at mechanisms by which this might come
about (with or without the help of the genes) before moving on
to summary arguments and the verdict.

198

PARADIGMS LOST

CONFLICTING RATIONALITIES AND THE
DILEMMA OF COOPERATION

In 1951 Merrill Flood of the RAND Corporation introduced one
of the most thought-provoking concepts in the history of strate-
gic thinking. His idea, later termed the Prisoner’s Dilemma by
Albert Tucker, cuts to the heart of an age-old question: How do
we balance individually selfish acts against the collective ratio-
nality of individual sacrifice for the sake of the common good? A
familiar example will illustrate the point.

In Puccini’s opera Tosca, Tosca’s lover has been condemned to
death, and the police chief Scarpia offers Tosca a deal. If Tosca
will bestow her sexual favors on him, Scarpia will spare her
lover’s life by instructing the firing squad to load their rifles
with blanks. Here both Tosca and Scarpia face the choice of ei-
ther keeping their part of the bargain or double-crossing the
other. Acting on the basis of what’s best for them as individuals,
both Tosca and Scarpia try a double cross. Tosca stabs Scarpia
as he is about to embrace her, while it turns out that Scarpia has
not given the order to the firing squad to use blanks. The di-
lemma is that this outcome, undesirable for both parties, could
have been avoided if they had trusted each other and acted not
as selfish individuals, but rather in their mutual interest.

The tragic fates of Tosca and Scarpia serve to characterize the
essential ingredients of a classical Prisoner’s Dilemma situation:
There are two parties, each of whom has the choice of either
cooperating (C) or defecting (D), i.e., acting either to sacrifice
their individual interests for the sake of a common good, or to
further their own selfish individual interests at the expense of
the other. In addition, there must be a payoff structure involv-
ing a temptation (T), the payoff received by defecting when the
other party cooperates; a reward (R), the payoff each party re-
ceives if they both cooperate; a punishment (P), the payoff they
each get if they both defect; a sucker’s payoff (S), which is the
amount received by the cooperating party when the other de-
fects. For the Prisoner’s Dilemma to arise, these payoffs must be
ordered largest to smallest in the following way: T > R > P >
S. To avoid getting locked into an out-of -phase cycle of mutual
defections and cooperations, there is the technical condition

IT'S IN THE GENES

199

(T + S)/2 < R. Under these conditions, let’s quickly analyze
the source of the dilemma faced by Tosca and Scarpia when con-
sidering their respective courses of action.

To make things concrete, let’s put in numerical values for the
payoffe in Tosca. Suppose they are T = 4, R = 3, P = 2, S = 1.
Tosca can then argue: If I defect and Scarpia cooperates, my
lover’s life will be saved and I won’t have to see Scarpia, yield-
ing a payoff to me of 4 units. But if I defect and Scarpia also
defects, then even if I do lose my lover, at least I won’t have to
give myself to that pig Scarpia and I’ll end up with 2 units. On
the other hand, if I trust Scarpia and he trusts me so that we
both cooperate, I’ll get 3 units, while if I trust him by cooperat-
ing and he double-crosses me and defects, then I’ll get only the
sucker’s payoff of 1 unit. So, all in all, by defecting I’m assured
of getting 2 units, whereas if I cooperate I can’t get any more
than 3 units and could end up with much less. Therefore, ration-
ally it’s in my best interest to defect. Of course, the situation is
perfectly symmetrical and Scarpia, being equally rational and
logical, comes to the same conclusion and also opts to defect. Re-
sult: Both Scarpia and Tosca end up with much less than they
could have had by showing a little mutual trust. In other words,
by employing individual rationality they sacrifice their collective
joint interests.

The relevance of the Prisoner’s Dilemma for sociobiology is
evident. The cornerstone of sociobiological reasoning is the claim
that human behavior patterns, including what look on the sur-
face like selfless acts of altruism, emerge out of genetically self-
ish actions. In the context of the Prisoner’s Dilemma, we can
translate this sociobiological thesis into the statement that the
individually rational act of defection will always be preferred to
the collectively rational choice of cooperation. Our question is
then : Can that situation ever lead to a population of cooperators ?
If there is no way for cooperative acts to emerge natur-
ally out of self-interest, it’s going to be very difficult for the
sociobiologists to support their case. Put in our earlier game-
theoretic terms, always to defect is an evolutionary stable strat-
egy, since players who deviate from this policy can never make
inroads against a population of defectors. Or can they? Are
there any situations in which a less cutthroat course of action
can ultimately establish a foothold in a population of defectors?
This was the Big Question that Robert Axelrod set out to an-

200

PARADIGMS LOST

swer in one of the most intriguing psychological experiments
carried out in recent years. The separate issues that Axelrod
wanted to address were: (1) How can cooperation get started at
all in a world of egoists? (2) Can individuals employing coopera-
tive strategies survive better than their uncooperative rivals?
(3) Which cooperative strategies will do best, and how will they
come to dominate?

Axelrod’s key observation was to note that while ALL D, the
strategy of always defecting, is uninvadable for a sequence of
Prisoner’s Dilemma interactions that is of known, fixed, and fi-
nite duration, there may be alternative ESS strategies if the
number of interactions is not known by both parties in advance.
So after having played a round of the Prisoner’s Dilemma, if
there is a nonzero chance that the game might continue for an-
other round, then maybe there is a nice strategy that is also
ESS. Here by “nice” we mean a strategy that would not be the
first to defect.

To test this idea, Axelrod invited a number of psychologists,
mathematicians, political scientists, and computer experts to
participate in a contest pitting different strategies against one
another in a computer tournament. The idea was for each partic-
ipant to supply what he or she considered to be the best strategy
for playing a sequence of Prisoner’s Dilemma interactions, with
the different strategies then competing against each other in a
round-robin tournament. Fourteen competitors sent in strate-
gies, which were in the form of computer programs. The ground
rules allowed the programs to make use of any information
about the past plays of the game. Furthermore, the programs
didn’t have to be deterministic, but were allowed to arrive at
their decision by some kind of randomizing device if the player
so desired. The only condition imposed was that the program ul-
timately come to a definite decision for each round of play: C or
D. In addition to the submitted strategies, Axelrod also included
the strategy RANDOM, which took the decision to cooperate or
defect by, in effect, flipping a coin. In the tournament itself,
every program was made to engage every other (including a
clone of itself) two hundred times, the entire experiment being
carried out five times in order to smooth out statistical fluctua-
tions in the random-number generator used for the nondetermin-
istic strategies.

The winning strategy turned out to be the simplest. This was

IT'S IN THE GENES

201

the three-line program describing the strategy TIT FOR TAT.
It was offered by Anatol Rapoport and consisted of the two
rules: (1) cooperate on the first encounter; (2) thereafter, do
whatever your opponent did on the previous round. That such a
simple, straightforward strategy could prevail against so many
seemingly far more complex and sophisticated rules for action
seems nothing short of miraculous. The central lesson of this
tournament was that in order for a strategy to succeed, it should
be both nice and forgiving, i.e., it should be willing both to initi-
ate and to reciprocate cooperation. Following a detailed analysis
of the tournament, Axelrod decided to hold a second tournament
to see if the lessons learned the first time around could be put
into practice to develop even more effective cooperative strate-
gies than TIT FOR TAT.

As prelude to the second tournament, Axelrod packaged up all
the information and results from the first tournament and sent
it to the various participants, asking them to submit revised
strategies. He also opened up the tournament to outsiders by
taking out ads in computer magazines, hoping to attract some
programming fanatics who might take the time to devise truly
ingenious strategies. Altogether Axelrod received sixty-two en-
tries from around the world, including one from the renowned
game theorist John Maynard Smith, mentioned earlier as the de-
veloper of the ideas of the evolutionary game and the ESS. The
winner? Again it was Rapoport with TIT FOR TAT! Even
against this supposedly much stronger field, Rapoport’s game-
theoretic version of the Golden Rule was the hands-down win-
ner. The general lesson that emerged from the second
tournament was that not only is it important to be nice and for-
giving, but it’s also important to be both provocable and recog-
nizable; i.e., you should get mad at defectors and retaliate
quickly but without being vindictive, and you should be
straightforward, avoiding the impression of being too complex.
After extensive study of the results, Axelrod summarized the
success of TIT FOR TAT in the following way:

TIT FOR TAT won the tournaments not by beating the other

player but by eliciting behavior from the other player that allowed

both to do well. ... So in a non-zero sum world, you do not have

to do better than the other player to do well for yourself. This is

202

PARADIGMS LOST

especially true when you are interacting with many different play-
ers. . . . The other’s success is virtually a pre-requisite for doing
well yourself.

So what are the implications of these results for sociobiology?

If we think of the total points amassed by a strategy during
the course of the tournament as its “fitness,” and if we interpret
“fitness” to mean “the number of progeny in the next genera-
tion,” and finally if we let “next generation” mean “next tour-
nament,” then what happens is that each tournament’s results
determine the environment for the next tournament. The fittest
strategies then become more heavily represented in the next
tournament. This interpretation leads to a kind of ecological ad-
aptation without evolution (since no new species come into exis-
tence). Sociobiologists can take heart in this sort of
interpretation of Axelrod’s experiments because they show that
it’s possible for phenotypically altruistic (cooperative) behavior
to emerge out of individually selfish motives. It’s important to
emphasize here, though, that these results say nothing about the
actual causal factors at work generating the individual motives.
They could be genetic, as hard-core sociobiologists would love to
argue, but there is nothing in Axelrod’s work to say that they
are. Nevertheless, the experiments do offer some support to the
sociobiological explanation of cooperative behavior by means of
reciprocal altruism.

Following his work on the evolution of cooperation, Axelrod
carried out another set of experiments that also give succor to
the sociobiologist’s claim for an evolutionary development of
standards of behavior, i.e., cultural norms. The basic idea was to
use a souped-up version of the Prisoner’s Dilemma in which the
players had the choice not only of cooperation or defection, but
also of punishing a defection or letting it pass. Players in the
Norms Game are characterized by two qualities: boldness (B),
which measures the risk they are willing to run in defecting; and
vengefulness (V), a measure of their inclination to punish defec-
tion. Strategies were assigned randomly to twenty players, with
the first round of play lasting until each player had had four
opportunities to defect. At the end of the first generation, a
strategy was given one offspring if its score was near average,
two offspring if its score was at least one standard deviation
above the mean, and no offspring if its score was more than one
standard deviation below the mean. Furthermore, Axelrod also

IT'S IN THE GENES

203

allowed for the emergence of new strategies through a process of
mutation in such a way that about one new strategy emerged in
each generation.

The results of the simulation showed that with enough time,
all populations would eventually converge to the collapse of the
norm, i.e., Y approaching zero. The problem appears to be that
the players lack sufficient incentive to punish the defectors: No-
body wants to play sheriff. As one way to enforce the norm, Ax-
elrod suggests a metanorm: direct vengeance not only against
those who defect, but also against those who refuse to punish
them. This is the kind of procedure we see in some totalitarian
countries where, when a citizen is accused by the authorities of
some real or imagined ideological transgression, others are called
upon to pile their own denunciations onto the back of the hapless
offender.

While these results are still in the preliminary stage, the Evo-
lution of Cooperation Game and the Norms Game both provide
some theoretical evidence in support of the idea that social be-
havior can emerge as the result of evolutionary processes involv-
ing individually selfish agents. Whether that selfishness is
programmed into the genes is anybody’s guess, but at least there
are no obvious game-theoretic barriers preventing it. Now let’s
leave the gaming arena and return to the courtroom to let the
respective sides make their final statements before we retire to
ponder the verdict.

SUMMARY ARGUMENTS

The claims and counterclaims have been flying fast and furious
throughout this chapter, so before trying to put them into some
sort of coherent order let’s restate the basic question to be de-
cided. The Prosecution’s contention is that:

The majority of human behavior patterns are strongly influenced by
the genes.

Note the emphasis here on the words “majority” and “in-
fluenced.” All that’s needed to make the sociobiologists’ case is to
agree that in the vast majority of situations, genetic makeup
plays a more important role than the environment in determin-
ing the way people act.

With this statement of the question in mind, let’s now turn to

204

PARADIGMS LOST

a tabular summary of what look to be the principal positions of
the contending parties. But before presenting the summaries
themselves, a few comments are in order:

1. Most of the advocates of sociobiology, as well as their critics,
expound a variety of arguments in support of their case. For
the sake of brevity, each table entry lists only a catchword or
two representative of the general position. No attempt has
been made to summarize all aspects of any promoter’s argu-
ment.

2. Oddly enough, the men most responsible for the theoretical
underpinnings of human sociobiology, Hamilton, Maynard
Smith, and Axelrod, appear to be pretty lukewarm, at best,
about the case for human sociobiology. In fact, Maynard
Smith, for one, categorically denies that there is any direct
contact between his work on game-theoretic models for animal
aggression and the behavior of humans. My suspicions are
that he does so to avoid being sucked into the bottomless pit
of ideological debate with supporters of the Boston Group.
Nevertheless, I have placed all this theoretical work under
the general heading of the Prosecution, since it acts to lend
more support to the adherents of sociobiology than to their
detractors.

3. Richard Dawkins appears in the curious position of support-
ing both the Prosecution and Defense on my lists, since his
original work on the selfish gene argues strongly for a genetic
basis to behavior, whereas his later discussion of cultural
memes is really more of a case for environment as being the
main motivator of human actions.

With these clarifications at hand, let’s examine the summaries of
the competing cases in Tables 3.1 and 3.2.

Prior to making tracks for the jury room, we must listen to
the judge’s instructions. In our assessment of the evidence, no
weight whatsoever is to be given to the wild-eyed, slightly hys-
terical political outburst made in open court by the Defense. Re-
gardless of any jury member’s personal sentiments, we’re in a
courtroom, not at a political pep rally, and the matter to be de-
cided here is a question of science, not politics. This should not
be taken to mean that the arguments of the Boston Group are
wrong, only that the political component of what they argue
should never have been heard, and wouldn’t have been if the
court had been quick enough to muzzle the Defense attorney

IT'S IN THE GENES

205

HUMAN BEHAVIOR IS PRIMARILY GENETIC!

PROMOTER

ARGUMENT

Lorenz innate aggression, group selection

Wilson, Barash genetic influence, multiplier effect

Dawkins selfish genes

Lumsden and Wilson coevolutionary circuit

Trivers reciprocal altruism

Hamilton

Maynard Smith

Axelrod

“theoretical support”

inclusive fitness, kin selection
evolutionary game, ESS
evolution of cooperation and norms

TABLE 3.1

. Summary arguments for the Prosecution

HUMAN

BEHAVIOR IS PRIMARILY
ENVIRONMENTAL!

PROMOTER

ARGUMENT

Boston Group

Schwartz

Sahlins

Gould

Dawkins

reification, no multiplier effect, unfalsifiability
evolutionary constraints
kin selection impossible

Just So stories

cultural memes

TABLE 3.2. Summary arguments for the Defense

properly. Despite Defense protests to the contrary, politics has
no place in the laboratory, however one might personally feel
about the nature of what’s being studied, or however much one
might wish for an experimental result to come out one way in-
stead of another. So put the political smokescreen and academic
vigilantism out of your mind when pondering your verdict.

BRINGING IN THE VERDICT

Of all the Big Questions dealt with in this book, I find the socio-
biology problem to be the most perplexing. Even after wading
through the pages of testimony and trying to filter out the nug-

206

PARADIGMS LOST

gets of real information from the fool’s gold of rhetoric and po-
litical bombast, when all is said and done I’m forced to take ref-
uge in that ancient Scottish verdict “not proven.”

In terms of hard evidence, I find that after we pass from the
few instances of behavioral maladies like schizophrenia that ap-
pear to be solidly founded cases of genetic causation, the tangi-
ble facts supporting the case of sociobiology fade away like a
trickle of water in the desert. On the other hand, the circum-
stantial evidence is impressive. Predictions of both animal and
human behavior made on the basis of sociobiological arguments
seem to be, for the most part, at least within the bounds of nor-
mal experimental error. And the idea of a smooth transition
from rather clear-cut evidence of genetically influenced behavior
in the animal kingdom to similar behavior patterns in humans if
appealing.

In many ways, my feeling is that the socio biologists have beei
a little too eager to promote their cause by calling forth evidence
of dubious validity, and neglecting some rather obvious alterna-
tive interpretations. All of this calls to mind the statement made
by Alexander Solzhenitsyn in his 1978 Harvard commencement
address when he noted that “hastiness and superficiality are the
psychic disease of the twentieth century.” It’s tempting to won-
der whether or not some of the Boston Group, as well as the
Wilson circle, were present in Cambridge that day when this
compact summary of many of their claims was expressed.

On the other side of the ledger, I also find the arguments
(scientific and philosophical, that is) of the Defense to be diffi-
cult to dismiss. For the most part, there really is no firm evi-
dence to support the kind of direct path from the genotype to
phenotypic behavior that the sociobiologists need to establish
their case. And it is true that the usual arguments leading from
behavior back to genetic causes leave a lot to be desired, both
scientifically and philosophically. Furthermore, Stephen Jay
Gould may have a valid point when he says that the genes have
given up their sovereignty over the major human behavior pat-
terns as a result of Homo sapiens ’s most distinguishing feature —
an extraordinarily large brain.

All things taken together, I feel a bit like the Dodo in Alice in
Wonderland when he announces the winners of the Caucus Race:
“Everybody has won, and all must have prizes.” Frankly, I can’t
for the life of me understand why, in the face of so much real

IT'S IN THE CENES

207

and circumstantial evidence supporting both sides of the debate,
the participants continue to cling so fiercely to what are basi-
cally either-or positions. To an outsider, it seems pretty clear
that most interesting human behavior patterns are brought
about through a complex combination of genetic and environ-
mental factors, and the real work should be addressed to investi-
gating these complicated webs of interconnection. To my mind it
seems a futile effort to try to disentangle the relative contribu-
tions of the genes and the environment, and even more futile to
dissipate energy on senseless political harangues about a distinc-
tion that is far more virtual than real. But before dismissing the
sociobiology debate in such a nonpartisan and cavalier manner, I
think it’s worth speculating a moment on some of the reasons
why sociobiology seems to touch such a sensitive nerve in the
psyches of the scientific community and the general public alike.

In my opinion, the root cause of the heated public and aca-
demic debates over the claims of the sociobiologists comes down
to only one thing — raw power. As the Boston Group has noted,
the reductionists flavor of sociobiology, and its implication that
human society is both inevitable and the result of adaptive pro-
cesses, holds great attraction for the John D. Rockefellers of the
world who wield power and want to justify their actions by an
appeal to that final authority, Nature. As the late ethologist
Niko Tinbergen expressed it:

It is tempting to ponder this over-emphasis on studies of causa-
tion. I believe that it is partly due to the fact that, as the develop-
ments of physics and chemistry have shown, knowledge of the
causes underlying natural events provides us with the power to
manipulate these events and bully them into subservience.

In short, sociobiology offers us a mystique of power.

An equally plausible and closely related reason why sociobiol-
ogy appears so compelling is offered by Barry Schwartz, when
he notes that we are living in a time when the pursuit of self-
interest in the free-market economy provides the primary meta-
phor for understanding social relations. As a result, our social
and cultural categories overlap with our economic ones. Conse-
quently sociobiology, with its explanatory structure based upon
the “economic accounting” of evolutionary biology, seems to
capture many of the most prominent features of modern life.
However, this “fortuitous” juxtaposition of economic, social,

208

PARADIGMS LOST

and biological principles is not a universal biological necessity.
Both economic and social situations can change; biological prin-
ciples cannot. Consequently, it may indeed be dangerous to
argue that the unbridled pursuit of selfish personal interest is a
part of basic human nature — just as Lewontin & Co. have been
claiming all along!

So we end up closing the book on the question of Nature ver-
sus nurture barely any farther along the road to an answer than
when we opened it. But there has always been one area in which
virtually everyone agrees that there is a biological substrate un-
derlying a uniquely human behavior: the capacity for semantic
language. As we might suspect by now, however, even this seem-
ingly clear-cut case is not without its competing factions. So as a
detailed case study of one small, but important, corner of the
social cum biological forest, let’s now move on to a consideration
of the problem of human language acquisition.

SPEAKING FOR
MYSELF

CLAIM:

HUMAN LANGUAGE CAPACITY STEMS
FROM A UNIQUE, INNATE PROPERTY OF
THE BRAIN

DUMB DOGS AND CLEVER HANS

Unlike Americans, Austrians have no prejudices or proscrip-
tions against allowing dogs into their restaurants. Consequently,
at the Kuchldragoner, a Viennese Beisl I frequent for lunch, the
house dogs, Chi-Chi and Isabella, routinely make their appear-
ance at my table to put in their claims for a sliver of schnitzel
with a low whine, a paw on my lap, or, in Isabella’s case, just the
dropping of her St. Bernard-like head onto the edge of my table.
Of course these two mutts don’t realize how crazy they are to
think that I’m ready to part with any of Frau Holzfeind’s
Specknockerl, Orammelknodel, or Schinkenfleckerl, so just as rou-

210

PARADIGMS LOST

tinely I try to convince them of the folly of their canine ways by
telling them, “Not now, Chi-Chi,” or “You’re so beautiful
today, Isabella, but this food isn’t for dogs,” or, if all else fails,
“Get lost, dog!” What could possibly make me think that these
hounds understand even one word of what I’m saying to them,
especially when I usually say it in a garbled, pidgin version of
their “native” German? In fact, after every such encounter I
end up feeling slightly foolish, often wondering who is really the
crazy one in our by-now-almost-ritualistic interactions.

Strangely enough, my experiences in linguistic communication
with Chi-Chi and Isabella seem to reflect an almost universal
human belief in the ability to communicate by speech, or at least
symbolic language, with the higher animals. As a small child I
was convinced our family dog was just brimming over with ideas
and plans he wanted to tell me about, and I remember asking my
mother why he couldn’t speak to me like my other playmates. To
her eternal credit, she responded with the commonsense answer
that perhaps he really didn’t have anything to say, or at least
not anything that would be of concern, or even comprehensible,
to any human. Later, though, I felt slightly better about my
“stupid” query when I read an article claiming that Alexander
Graham Bell had tried teaching a dog to speak by training it to
growl at a constant level while he manipulated its jaw muscles
and throat to get it to produce various sounds. About the best
the poor pooch could come up with was something that sounded
a lot like ah oo yow grrr, a pretty poor imitation of “Let me out
of here,” at which point Bell wisely returned to his work on the
telephone.

Undeterred by failures of this sort actually to speak with ani-
mals, around the turn of the century a retired German school-
teacher named Wilhelm von Os ten acquired a bit of a local
reputation in Berlin by displaying his horse Hans, which he
claimed could solve problems in arithmetic. The question would
be posed, “Hans, how much is three plus five?” — at which point
Hans would start tapping his hoof, stopping after the eighth
tap. It makes one wonder what Hans’s response would have been
if he were asked to extract the square root of 7r! Unfortunately
for both von Osten and Hans, the psychologist Oskar Pfungst
made a detailed investigation of the “Clever Hans” phenome-
non, conclusively showing that Hans was more of a showman
(showhorse?) than an accountant, taking unconscious cues from

SPEAKING FOR MYSELF

211

his owner during the course of his demonstrations. In fact,
Pfungst demonstrated that Hans had the uncanny ability to de-
tect head movements of as little as one fifth of a millimeter,
thereby being able to “read” the slight, but unconscious, move-
ment in his trainer’s head when he came to the correct number
of taps for a given computation.

The most recent manifestation of the human psychological
need to talk with the animals has been the spate of experiments
by John Lilly, David Premack, Allen and Beatrice Gardner,
Herbert Terrace, and others aimed at communication with dol-
phins, gorillas, and chimpanzees by sign language, manipulation
of colored chips, and other such devices. While virtually every-
one agrees that some kind of interspecies communication has
taken place in these efforts, the bottom line seems to be that
whatever kind of communication it is, it’s not what we think of
when we consider human linguistic interaction. The key question
that emerges from this result is why anyone would seriously en-
tertain the notion that chimps, whales, or apes could be taught
to communicate using the same principles that underlie human
language in the first place. The answer involves a brief consider-
ation of the historical origin of language.

The number of different theories of how human language
originated seems to be about equal to the number of investiga-
tors of the topic, ranging from “bowwow theories” claiming that
language emerged from imitations of the sounds of nature with
onomatopoeic words, to “singsong theories” claiming that speech
arose from the love songs and rhythmic chants of primeval Lo-
tharios. Following an explosion of such wild speculation, in 1886
the Linguistic Society of Paris issued a resolution “outlawing”
any more papers concerned with the origin of language. The ban
was upheld in 1911 by the Philosophical Society of London but,
regrettably, it does not seem to have stanched the flow of specu-
lation on the topic. The most sober guesses today argue for the
origin of language as an evolutionarily advantageous trait en-
abling primitive man to communicate more effectively in groups
for hunting, socializing, and defense. Whatever the actual rea-
son for its origin (and it’s likely to be a combination of several
causes), human language is almost universally considered to
have evolved out of more primitive levels of brain neurophysi-
ology and body structure.

212

PARADIGMS LOST

If we accept the picture of human language as having its ori-
gin in an evolutionary outgrowth of body and brain develop-
ment, what could be more natural than to ask: What was the
first human language? According to the second book of Herodo-
tus’s History, credit for the initial experiment on the matter
goes to the Egyptian pharaoh Psammetichus, who around
twenty-five hundred years ago arranged to have two infants
raised in a “linguistic deprivation tank” under the assumption
that whatever their first word turned out to be, it would be from
mankind’s true “source” language. Herodotus reports that the
first word uttered was bekos, the word for “bread” in Phrygian,
a language then spoken in the northwestern corner of what is
now Turkey. Thus, Psammetichus concluded that Phrygian was
mankind’s original tongue. In good scientific fashion, the mon-
archs James IY of Scotland and Frederick II of Hohenstaufen
both repeated the pharaoh’s experiment, with the outcome that
James’s test subject “spak guid Ebrew.” Unfortunately, Fred-
erick’s subjects died, perhaps from loneliness, before having the
chance to utter even a single word. The sum total of all these
experiments is that we have no more real information about the
original language of man than did Psammetichus, but the exer-
cises do allow us to contrast in sharp colors the lines of research
that linguists have taken since the pharaoh’s time.

In terms of philosophical endearment, linguistic research can
be divided into two main camps: the empiricists and the rational-
ists. The thesis of the first group is that the only way to under-
stand human language is by actual observation. Go into the field
with your tape recorder and notebook, gather several hundred
hours of actual speech in a variety of situations, then analyze
the data to extract the linguistic patterns present in a particular
speech community. The rationalists, on the other hand, hold to
the view that there’s much more to language than just data;
there is an innate knowledge of linguistic structure that is part
of the genetic birthright of any normal human child, and in
order to understand language it’s necessary to take this innate
knowledge into account. In the more biological terms used in the
last chapter, the empiricist stance corresponds to the position
that language is basically determined environmentally, while the
rationalists cling to the position that it’s principally in the
genes. To sort out the competing claims, a short tour of twen-
tieth-century linguistics research will prove helpful.

SPEAKING FOR MYSELF

213

VERBAL BOTANY
AND UNIVERSAL GRAMMAR

According to a count by that bastion of linguistic conservatism
the Academie Frangaise, there are currently 2,796 separate dia-
lects spoken on Earth, a number no doubt contributing to the
old joke that a language is nothing but a dialect with an army
and a navy. When contemplating this enormous variety of lan-
guages, especially in light of my own anemic linguistic talents, I
become ever more sympathetic to the view of Naguib Mahfouz,
1988 Nobel laureate in literature and the first Arab writer ever
to be so honored, when he stated that it would be much better for
culture and humanity if all writers wrote in the same language.
But, alas, both Mahfouz and I seem to be stuck with the Acade-
mic’s list of 2,796. Every one of these known languages shares
the following features:

1. Formation of a large number of meaningful symbols (words)
from a small set of basic sounds (phonemes).

2. Formation of an unlimited number of sentences by logically
combining words using a finite number of grammatical rules.

3. The sentences are used for socialized actions.

4. Any normal child has the ability to learn to speak the lan-
guage.

By way of contrast, no known system of animal communication
shares all of these characteristics. For example, the dance of the
honeybees doesn’t involve symbols or sentences, nor is it learned.
And the chimpanzees don’t form structured sentences either.
The science of linguistics has arisen over the past century or so
with the goal of studying the properties of these 2,796-plus ways
of human communication.

It seems that all linguistic researchers agree on what consti-
tute the basics: A language is composed of a set of meaningful
sentences, each composed of a set of words, each of which is in
turn formed phonetically out of a set of elementary sounds (pho-
nemes) like [f] and [v] in “fine” and “vine,” and semantically
from a collection of “atoms” of meaning (morphemes) such as
[im] and [possible] in the two-morpheme word “impossible.”
Further, for each language there is a grammar consisting of the

214

PARADIGMS LOST

rules determining the allowable ways in which the words can be
combined to form sentences (syntax), as well as the manner in
which those sentences are to be understood ( semantics ) and pro-
nounced (phonetics). Thus, the grammar specifies the totality of
linguistic rules necessary for using the language. (This technical
use of the term grammar is not quite the same as the grammar
most of us struggled with while parsing sentences in “grammar
school.” The differences will be made explicit later.) The overall
goal of virtually every linguistic researcher is somehow to spec-
ify explicitly the grammar characterizing any given language.
The fun begins when it comes to the question of just how we
should go about attaining this state of linguistic bliss.

According to one school of linguistic thought, the localists, the
most interesting aspects of languages are the ways in which they
differ. Accordingly, localists follow the empiricist path to under-
standing grammar, tending to emphasize the collection and anal-
ysis of field data on exotic languages like Philippine Tagalog,
Haitian Creole, or perhaps the Mandingo tongue of West
Africa. The approach of the localists is to start with a descrip-
tion of the elements in a language (the phonemes and mor-
phemes), and build up to the more complex elements (the
sentences). The underlying localist belief is that through acquisi-
tion of enough data, the patterns characterizing the grammar of
the language will slowly but surely emerge.

Following what seemed to be the eminently sensible approach
of gathering data before engaging in any theorizing, the localists
were first off the mark in the linguistics derby with the work of
Ferdinand de Saussure in Geneva. Saussure focused on lan-
guage as a system and tried to describe that system as a collec-
tion of interdependent parts deriving their significance from the
system as a whole. A second stream of localist thought was initi-
ated by the German linguist Franz Boas, who advocated what
amounts to an anthropological approach to analyzing the speech
patterns of living languages. Later the influential American lin-
guist Leonard Bloomfield brought these ideas to the forefront of
American linguistic research, developing localist methods and
notations for the study of exotic, unusual languages. Bloom-
field’s influential 1933 book Language dominated American lin-
guistic thinking for over twenty years. Its overriding theme was
the emphasis on objective methods of verification and precise
techniques of discovery, as well as a refusal to admit discussion

SPEAKING FOR MYSELF

215

of meaning or mental entities or any other kind of unobservable
features in the mind of the speaker. It’s against this backdrop of
behavioral psychology and logical positivism that the globalist
school of linguistic research emerged.

In direct contrast to the localists’ position, the globalist creed
is that the important parts of languages are their similarities,
not their differences. And the best way to study these similari-
ties is by admitting the discussion of possibly unobservable men-
tal structures giving rise to linguistic universals. Accordingly,
the globalists emphasize a top-down research program focusing
upon the abstract, syntactic structure of language per se, plac-
ing far less weight on the peculiarities associated with the con-
crete surface structure of any particular spoken language.

The modern era of globalist thought in linguistics research
was dramatically ushered in with the publication of Noam
Chomsky’s Syntactic Structures in 1957. This electrifying event
shifted the focus of linguistics virtually overnight from the ob-
servation and classification of the localists’ “verbal botany” to a
new vision of language as phonetics and semantics superimposed
upon an underlying core of pure syntax. The principal goal of
research now was to identify this core universal grammar from
which all human languages get their start. In the globalist view,
the universal grammar is something that is biologically present
in the mind of all normal children as part of their genetic birth-
right. Since we’ll consider Chomsky’s program in painstaking
detail later, it suffices for the moment to note that in addition to
proposing the idea of a universal grammar forming the abstract
structure upon which all languages are built, Chomsky also put
forth in Syntactic Structures the radical notion that the grammar
of each language must be generative in the sense that it must be a
set of rules capable of “generating” all the well-formed (i.e.,
grammatical) sentences of the language and none of the ill-
formed ones.

In addition to providing a set of formal tools and a theoretical
framework for investigating the abstract properties of lan-
guages, Chomsky’s work had the far-reaching effect of totally
reorienting the primary direction of linguistic research. For the
globalists, with their preoccupation with linguistic universals,
the main focus now became not the speech patterns of adult
speakers, but a deeper understanding of the process by which

216

PARADIGMS LOST

children come to learn their native language. In fact, it’s fair to
say that the globalists’ program is directed toward answering

THE CENTRAL PROBLEM OF MODERN LINGUISTICS

How do children acquire the ability to speak their native language f

Upon first hearing this question, most people would dismiss
the problem of language acquisition as no problem at all, saying
that children obviously learn to speak by listening to their par-
ents and older playmates. Unfortunately, this commonsense re-
sponse just doesn’t stand up to the test of observation, the main
obstacle being what is often called the “poverty of the stimulus”
problem. Since it forms one of the pivotal points of this chap-
ter’s debate, let’s look at what the stimulus deficiency problem
means in somewhat greater detail.

In broad terms, “the poverty of the stimulus” refers to the
fact that, during the linguistically formative years, the child is
not exposed to enough language to account for the linguistic ca-
pability displayed by a normal six-year-old. In short, children’s
ability to use their native language is vastly underdetermined by
the data. There are several aspects to this underdetermination,
each of which strongly suggests the need for something beyond
mere exposure to account for the phenomenon of language ac-
quisition.

First of all, the speech the child hears doesn’t always consist
of well-formed, complete sentences, but includes ill-formed sen-
tences, partial statements, slips of the tongue, and other incom-
plete and/or ungrammatical utterances. Furthermore, children
encounter only a finite range of expressions, yet come to be able
to deal with an infinite spectrum of novel sentences going
far beyond what they have ever heard before. Somehow the
child acquires schemes for generating potentially infinite sen-
tences such as “This is the dog that chased the cat that ate the
mouse . . .” (relativization) or “Susan went home, and Jerry
and Jane went out, and Carl slept . . .” (coordination) or “You
heard that John asked me to tell Sam that he saw the house
. . .” (subordination). You’ve probably never seen these sen-
tences before and, in fact, it’s likely that they’ve never before
been written in any book until I made them up today. Neverthe-
less, you immediately understand the structure and meaning of

SPEAKING FOR MYSELF

217

them and, what’s more important, so does your five-year-old
child. So it can’t possibly be the case that children learn their
native language solely by imitation of what they hear. Finally,
children come to know things subconsciously about their lan-
guage for which there is no direct evidence in the data to which
they are exposed. For example, children are not systematically
informed that some hypothetically possible sentences do not in
fact occur, or that a given phrase such as “I like her cooking” is
ambiguous.

To summarize the relevant facts about language acquisition:

1. The child masters a rich system of knowledge without signif-
icant instruction.

2. This is carried out despite the poverty of the stimulus.

3. The process takes place most rapidly between the ages of two

and three.

4. Normal human children are able to master any human lan-
guage to which they are exposed in infancy.

An integral part of the globalist program is the assertion that
any theory of human languages must provide an explanation for
the above empirical facts, and that such an explanation will
never be forthcoming by following the butterfly-collecting path
of the localists.

By and large Chomsky’s revolution has driven the verbal
botanists underground, the main fireworks in contemporary lin-
guistic research now centering upon his program. As we’ll soon
see, this program has many parts, some of which are technical,
others psychological and philosophical. Moreover, Chomsky’s
own position has shifted somewhat over the past thirty years, so
the program presented in Syntactic Structures no longer repre-
sents a totally faithful account of his vision. Nevertheless, cer-
tain key points have remained invariant, one of them being the
dogmatic claim that the human language acquisition capability
is attributable to a unique, genetically programmed part of the
brain. It is on this point of innateness that many psychologists
as well as linguists balk, and it is also at this point that
Chomsky’s views on linguistics most dramatically intersect with
other philosophical, psychological, and neurophysiological ques-
tions of brain, mind, body, and thought. Thus, our debate in this
chapter focuses on the problem of language acquisition as a ve-
hicle with which we can enter into some of the central themes of

218

PARADIGMS LOST

what has come to be known as cognitive science. As is the cus-
tom, we begin the trial with the Prosecution.

THE NOAM OP CAMBRIDGE

During a stormy crossing of the Atlantic in 1953 on an old tub
that had been salvaged after being sunk by the Germans during
the war, a seasick, twenty-five-year-old graduate student from
Philadelphia had an idea that was to initiate a bona fide Kuhn-
ian revolution in the way we think about language. That youth-
ful traveler was Noam Chomsky, and the idea he had was that
the peculiarities of the biological structure of the human brain
play the essential role in the ability humans have to communi-
cate by means of language. As Chomsky now tells it, “I remem-
ber exactly the moment when I finally felt convinced,” and at
that moment he set out on a course of study emphasizing the key
role of the mind and its mechanisms for making human language
possible. In Chomsky’s terms, the brain contains a genetically
programmed “language organ” enabling human children to mas-
ter their mother tongue with virtually no training or effort. Yet
at the same time this organ defines and circumscribes the bound-
aries of all human languages, specifying what is and isn’t possi-
ble by way of human linguistic communication.

As we’ve seen, prior to Chomsky’s brainstorm, linguists didn’t
think of brain structure as playing any significant role in shap-
ing human language. They thought of the mind as a tabula rasa
capable of learning any kind of language whatsoever, and con-
centrated on isolating “discovery procedures” that would objec-
tively describe the grammar of any human language. Although
starting out himself on this same structural linguistic path while
a student of Zellig Harris at the University of Pennsylvania,
Chomsky, after several years of effort, came to the conclusion
that some radical new notion was needed to understand the na-
ture of human language. His insight on that fateful sea voyage
was twofold: (1) recognition that the actual structure of the
brain was crucial for explaining human language ability, and
(2) recognition that the usual definition of grammar needed to
be expanded to include all the rules and elements of language
that children assimilate as they learn to speak and understand,
as well as the linguist’s theory of what goes on in the speaker’s
and hearer’s brains.

SPEAKING FOR MYSELF

219

In Chomsky’s view, heredity must play an overwhelmingly im-
portant role in language because there is no other way to ac-
count for the facts noted earlier surrounding childhood
language acquisition. In this genetically dominated picture,
there are special neural circuits for the representation and use
of language that interact with the child’s linguistic environment,
eventually evolving into a neurophysiological pattern specifying
the grammar of the language that the child ends up speaking.
According to this scenario, language growth is just one in the
long series of genetically programmed changes a child goes
through while maturing. Thus, just as the child is predeter-
mined to pass through puberty or lose baby teeth in a certain
genetically programmed manner, so it is with language as well,
with the crucial changes taking place beginning around the age
of two years and ending at about the onset of puberty. Among
other things, this explanation of language accounts for the way
children pick up a language as easily as the rest of us pick up a
cold, while it’s so painfully difficult for most adults to learn to
speak in a foreign tongue.

The key element in Chomsky’s biologically based theory of
language is the idea of a universal grammar, which we briefly
discussed in the last section. Upon first hearing this term, I con-
jured up the image of somehow taking all 2,796 human lan-
guages and throwing them together into a pot and boiling the
whole mess down to its distilled essence, the residue remaining
being the universal grammar. In some ways this image isn’t too
far off the mark, as the universal grammar does represent the
totality of all the immutable principles of language that Nature
builds into the language organ. But a better metaphor for the
universal grammar would be that of a not completely specified
electrical circuit. Figure 4.1 displays a simple passive electrical
circuit consisting of a single resistor R, a capacitor C, and an
inductor L, together with a voltage source. The way the circuit
transforms a signal at its input terminal into an output pattern
is governed by two factors: the way the circuit elements are con-
nected, and the actual numerical values of R, L, and C. In the
figure we see the circuit wired in two different ways, a pure se-
ries connection (a) and a series-parallel connection ( b ), with
other combinations also possible. In linguistic terms, we could
think of the input signal as corresponding to the child’s linguis-
tic experience, i.e., the external environment, and the output as
being the language produced by the mental language organ. The

220

PARADIGMS LOST

o sAAA

\SlSlSU If

-©

FIGURE 4.1. Passive electrical circuits: (a) series, (b) series-parallel

organ itself, the universal grammar, is represented by the com-
ponents of the electrical circuit, together with the manner in
which they are connected ( not including their actual numerical
values). Thus, the universal grammar is a set of preprogrammed
subsystems of the circuit (the resistor, capacitor, and inductor
together with their wiring pattern), but subsystems that are not
programmed down to the last detail (the actual values of R, L,
and C ).

The idea is that the linguistic input the child experiences re-
sults in setting the parameter values of the universal grammar,
thereby turning the language circuits in the brain into a lan-
guage device suitable for producing and understanding the spe-
cific language corresponding to the particular parameter
settings. The wiring pattern of the resistor, the capacitor, and
the inductor corresponds to the biologically innate universal
grammar. Fixing the actual values of R, L, and C corresponds

SPEAKING FOR MYSELF

221

to setting the “switches” of the universal grammar to produce a
specific human language.

The universal grammar characterizes the abstract syntax of
language, independent of the peculiarities and idiosyncrasies
present in a given human speech community. Thus heredity pro-
vides the basic outline common to every language, the child’s
linguistic environment then filling in the details pertinent to the
language being learned. While the universal grammar allows the
learning of any human language, it imposes rather narrow limits
on the possible ways that the rules governing each of its subsys-
tems can interact. For instance, languages like Italian have what
is called the null subject option, allowing statements such as
“went” instead of “he went” or “she went.” English has passed
up this option. It is the collection of such options that consti-
tutes the boundaries of the universal grammar, just as it is the
collection of choices for the values of R, L, and C that consti-
tutes what can be done with the “universal passive circuit.” But
it should be noted that the grammatical options, unlike their
electrical counterparts, cannot be chosen freely; the grammatical
options are interconnected, residing in a hierarchy where a
choice at one level constrains what can be done farther on down
the line. It’s also of critical importance to observe that the uni-
versal grammar says nothing about the lexical facts of a lan-
guage, but only about the form of the lexicon. Thus all
considerations of word categories such as nouns and verbs are
absent from the universal grammar. But it does contain princi-
ples about the assignment of semantic roles, cases, and so forth.

It’s now easy for us to see why Chomsky’s vision of universal
grammars and biologically based language organs resulted in a
total redirection of the lines of linguistic research. Rather than
trying to construct the grammar of an actual language by as-
sembling it piece by piece from individual observations, the
Chomsky program turned the process totally upside down and
started with the assumption that a universal grammar exists,
with the parameters of the grammar fixed but unknown. The
game is then to try to deduce these parameter settings from the
observed grammar of the actual human language under investi-
gation. So we see that the ultimate goal is the same — the charac-
terization of the grammar of a real language — but the path is
radically different: top-down instead of bottom-up. Moreover,
the important questions now become matters involving the deter-

222

PARADIGMS LOST

mination of the properties of the universal grammar rather than
matters of collection and analysis of field data.

It’s not surprising that Chomsky, who diametrically opposed
their research strategies, didn’t exactly capture the fancy of the
localists, who reigned over the American linguistic community in
the 1950s. As with most revolutionary insights, Chomsky’s ini-
tial attempts to publish his ideas, both in the form of a summary
article submitted to the prestigious journal Word and as a man-
uscript emerging from his Ph.D. work, met with a steady stream
of rejections, mostly under the impetus of reviews from the
Bloomfieldian old guard. Happily, through the influence of
Roman Jakobson, one of the founders of the Prague school of
linguistics and an influential member of the American linguistic
community, a drastically slimmed-down version of Chomsky’s
vision was finally published by the small Dutch house Mouton
under the title Syntactic Structures in 1957. Following a very fa-
vorable review of the book in the widely read journal Language,
everything, as they say, hit the fan. Chomsky was immediately
catapulted into a professorship at MIT and a position at the cen-
ter of the academic and intellectual stage, both of which he occu-
pies to this day.

Clearly the kind of furor that Chomsky stirred up in the lin-
guistics world, with his speculations about how language acqui-
sition works using an unobservable mental organ and a
universal grammar, didn’t arise solely because a new Ph.D. was
putting forward a few wild fantasies. There was much more to it
than that. So let’s go exploring and look more deeply at the main
features of this revolutionary research program and the array
of technical ideas Chomsky introduced to support it.

As with most great intellectual breakthroughs, Chomsky’s
started when he looked at an everyday occurrence from a new
point of view. He began by noting that there are certain proper-
ties of sentences that people intuitively know, but that can be
explained only by employing the kind of deep principles of lan-
guage that are known explicitly to linguists alone. The classic
illustration is his sentence “Colorless green ideas sleep furi-
ously,” which every speaker of English perceives as meaning-
less, yet perfectly correct grammatically. Somehow there is an
intuitive understanding at work here that tells us that the for-
mal structure of the sentence (its syntax) is fine, but that the
sentence is all form and no content. The point is that syntactic

SPEAKING FOR MYSELF

223

categories can be defined independently of meaning. Lewis Car-
roll’s “Jabberwocky” poem is another classic example of this
phenomenon (as are a depressing number of homework exercises
from my students). In other words, the syntactic rules used to
form the sentence exist independently of its semantic content.
Chomsky boldly asserts that what’s important about language is
the understanding of these syntactic rules, and that such under-
standing will never come about from looking in an inductive
fashion at just the utterances themselves. Instead, it’s necessary
to work deductively from a postulated set of rules, i.e., the uni-
versal grammar. Part of Chomsky’s insight was also to recog-
nize that he could cut the overall problem of grammar
identification down to digestible proportions by invoking the
simplifying assumptions that syntax could be studied indepen-
dently of other aspects of language, and that linguistics could be
pursued independently of other areas of cognitive science like
psychology, neurophysiology, and logic. Later in the chapter
we’ll turn to a consideration of how well these simplifications
hold up under detailed scrutiny.

Recall that for Chomsky the goal of linguistic research was to
provide a framework for characterizing the rules that specify
which sentences of a language are syntactically correct and which
are not, i.e., to specify a grammar for the language that correctly
distinguishes the “good” sentences in the language from the
“bad” ones. In actual practice, Chomsky demanded much more
than this. A key ingredient in his program was the claim that
every such grammar would not only have to be a decision proce-
dure for grammatical correctness, but it would also have to be a
generative grammar. That is, the grammar would have to have the
capability actually to generate all the well-formed sentences of
the language and none of the ill-formed ones. Chomsky’s ap-
proach was to present a sequence of increasingly powerful gram-
mars, and to argue that only the last grammar on his list would
serve as a viable candidate for the structure of a universal
grammar. The three principal delicacies on this Chomskian
menu are finite-state, phrase-structure, and transformational gram-
mars. Let’s look briefly at each of them in turn.

FINITE-STATE GRAMMARS

Using such a grammar, sentences are generated by a series of
choices, one after another. First a word is chosen from a set of

224

PARADIGMS LOST

possible words, where the actual choice is dictated by a random
selection weighted according to a given probability distribution.
Then the second word is chosen, also according to a probabilistic
weighting of choices, with a third word selected in the same
manner and so on. Mathematically, this kind of sequential selec-
tion is termed a finite-state Markov process when there are only a
finite number of sets from which the words are chosen. It is
called a first-order process if the probabilities affecting the
choice of a word depend only upon the preceding word, a second-
order process if the probabilities depend upon the two preceding
words, and an wth-order process if they depend upon the w
preceding words, n being finite. An illustration of such a process
is shown in Figure 4.2.

In the grammar of Figure 4.2, the node I is the starting node,
while node T represents a termination node. In this elementary
setup, there are only two grammatically correct sentences speci-
fied by the grammar: “The girl spoke” and “The men work,”
with the single probabilistic choice being made following the
word “the.” This primitive finite-state grammar starts in the
initial state I, then moves to the state “the” with probability
one, complete certainty. From the state “the,” the grammar can
move to either the state “girl” or the state “men,” each with a
probability that may depend upon the previous state, “the.”
And so it goes in this manner, with the grammar eventually gen-
erating the two grammatically correct sentences of this super-
primitive “dialect” of English. It’s clear from this simple case
that a finite-state grammar without feedback loops can generate
at most a finite number of sentences, hence could never serve to
characterize the grammar of any human language. But what if
we allowed loops? In that case it should at least be possible to
produce sentences that are, in principle, of infinite length.
Would this be enough to say that such a grammar is a viable

girl spoke

FIGURE 4.2. A simple finite-state grammar

SPEAKING FOR MYSELF

225

candidate for describing the rules of some human speech com-
munity?

Figure 4.3 displays a finite-state grammar with loops. With
such a grammar we can clearly generate sentences of infinite
length. Nevertheless, it is still a finite-state device. Why? Be-
cause whatever state it’s in (whatever box it’s approaching), no
matter what its previous states have been, the device still pro-
ceeds in exactly the same way. It has no way of “remembering”
how many times it has visited a given state, since if it did such
information would be part of its state and it would no longer be
a finite-state device, as such loopings can, in principle, be carried
out an infinite number of times.

In Syntactic Structures, Chomsky proved that a grammar based
upon such a device cannot possibly characterize human lan-
guages since it’s inherently incapable of accounting for nonadja-
cent dependencies. For example, the relationship between “toys”
and “are” in the sentence “The toys in the store . . . are funny,”
where “. . .” represents an indefinite amount of material, cannot
be handled by any kind of rule coming out of a finite-state gram-
mar. Thus did the idea of a finite-state grammar fall by the way-
side in Chomsky’s search for the right abstract structure for his
universal grammar.

PHRASE STRUCTURE GRAMMARS

With this type of grammar, we return to elementary-school days
and the standard exercise of diagramming sentences by follow-
ing a set of what linguists term phrase structure rules. Grade-
school days notwithstanding, these rules are easy enough to
understand, consisting of a set of statements such as:

• A sentence S consists of of a noun phrase (NP) followed by a
verb phrase ( VP).

• A noun phrase can consist of an article ( Art ) and a noun (IV).

• A verb phrase can consist of an auxiliary phrase (Aux ), a verb
( V), and another noun phrase.

These rules can be represented compactly in “rewrite” form as

S - NP + VP
NP - Art + N
VP - Aux + V + NP

226

PARADIGMS LOST

FIGURE 4.3. A finite-state grammar with hops

SPEAKING FOR MYSELF

227

and so on. Thus, the grammar for a simple fragment of English
might consist of rules of the above sort, which would be used to
decompose a sentence into its atomic constituents. As an exam-
ple, consider the decomposition:

Art N

The dog will eat the food

This derivation is Chomsky’s description of the syntactic struc-
ture of the sentence “The dog will eat the food.” It consists of a
set of phrase structure rules, together with the lexicon stating
what is to count as a noun, verb, auxiliary, and so forth. The
“tree” sitting above the target sentence is often called a phrase
marker, and is the structure resulting from applying only the
phrase structure rules to the original sentence and inserting the
lexical information.

Let’s note in passing that we now see the difference between
what is commonly termed grammar in everyday speech and what
linguists think of as a grammar. Everyday grammar is nothing
more than the rules of a phrase structure grammar, whereas a
linguist’s grammar is something far more general — any set of
rules that is capable of generating the correct sentences and only
the correct sentences of a human language, together with the
rules for their interpretation and enunciation. With this not so
crucial point settled, let’s get back to Chomsky.

Rather soon Chomsky saw that phrase structure grammars
with their single set of rules alone might succeed in characteriz-
ing the proper sentences of some language, but they could do so
only with undo complication and at the expense of introducing
an unwieldy number of rules. Moreover, with such a grammar
there is no natural way to describe ambiguous sentences such as
“I like her cooking.” Phrase structure rules would provide only
one parsing or “diagramming” for this sentence, but since the
sentence is syntactically ambiguous, any decent theory of gram-
mar should account for this fact by providing a number of syn-
tactic derivations and descriptions. In addition, surface
differences often conceal underlying similarities, as in the sen-

228

PARADIGMS LOST

tences “The dog will eat the food” and “The food will be eaten
by the dog.” These sentences mean the same thing, the only dif-
ference being that one is in the passive voice while the other is in
the active voice. Phrase structure grammars give us no way to
represent this similarity. With such examples and problems in
mind, Chomsky went on to propose his final level of grammatical
sophistication.

TRANSFORMATIONAL GRAMMARS

Since a single level- of operations, the phrase structure rules,
isn’t enough to pin down the richness of human language, the
“obvious” next step is to introduce a second level. Or so thought
Chomsky when he proposed the idea of a type of transformation
rule that would act to transform not grammatical categories like
noun phrases, verb phrases, and the like, but rather the entire
phrase marker itself. Thus, in transformational grammars there
is a second set of rules that act on one phrase marker to trans-
form it into another by actions such as moving elements around,
adding elements, deleting elements, and so on. For instance, we
can use Chomsky’s transformational rules to display the
similarity between the active and passive voices by showing how
the active and passive can be converted into each other by trans-
formations of the corresponding underlying phrase markers.

To illustrate the way such a grammar works, consider the
phrase marker below, diagramming the sentence “All the boys
might have gone with their parents,” where the element Q repre-
sents a quantifier, P is a preposition, and PP is a prepositional
phrase.

Q Art N

all the boys

V VP

/ \

V VP

/ \

V PP

/ N
P NP

Art

might have gone with their

parents

In a sentence of the above sort, the quantifier “all” has consid-
erable flexibility as it can occur once for each subject NP. It

SPEAKING FOR MYSELF

229

appears that no simple phrase-structure rule is capable of pro-
ducing the full range of grammatical sentences that can come
out of the above initial phrase marker without producing un-
grammatical sentences, too. For instance, the rule “Allow an op-
tional occurrence of Q before each of the verbs” will correctly
produce a sentence like “The boys all might have gone with their
parents,” but it will also produce the ungrammatical “All the
boys all might have gone with their parents,” which comes from
a double usage of Q, an action admissible by the above rule.

What’s needed here is a new kind of rule that can scan an
entire phrase-structure tree at one glance, and instead of just
marking a tree as grammatical or not, is capable of actually
transforming it into a new configuration. In our example, the
kind of rule that’s needed is one like “Detach Q from within an
NP and move it to the left of any verb in the structure.” Appli-
cation of this rule to the foregoing tree yields:

Art

.VP

. VP.

. VP

Art

the

boys all might have gone with their

parents

Corresponding to the phrase structure rules and the transfor-
mational rules, respectively, are two subsystems constituting the
syntax of a language: a base subsystem and a transformational
component. The base subsystem contains the phrase structure
rules, which, in Chomsky’s terminology, determine the deep
structure of every sentence. The transformational component of
the grammar then transforms the deep structure into its surface
structure, the level at which the phonetic components of the lan-
guage take over to give the sentence a phonological structure. In
the early versions of Chomsky’s theory, the deep structure was
also employed without benefit of transformation as input to the
semantic component of the language’s grammar in order to give
the sentence its actual meaning. In later years, this firm distinc-
tion between syntax and semantics has been blurred somewhat,

230

PARADIGMS LOST

even by Chomsky himself, although the relative importance of
the deep structure for semantics is still a matter of hot debate.

It should be noted that the term deep in this Chomskian con-
text has no bearing on matters of profundity, but signifies only
the “hidden,” purely abstract, syntactic structure of the sen-
tence. Using his transformational rules, Chomsky was able to
show that ambiguous sentences such as “I like her cooking”
could be given a single surface structure from several deep
structures, while semantically equivalent sentences of the sort
involving just a change from active to passive voice could have
different surface structures emerging from the same deep struc-
ture.

This summarizes the Chomsky revolution in linguistics in a
nutshell. The whole program is set out in Table 4.1.

Another way of looking at Chomsky’s work is to examine its
implications not only for linguistics, but also for psychology and
the philosophy of mind. Chomsky’s principal claims in these
areas are:

Psychology

• It makes sense to speak of abstract, possibly
unobservable, mental entities.

• One of these mental “organs” is designed
specifically for language.

• This language organ, the so-called universal
grammar, is genetically determined.

Linguistics

• To discover grammars, studies must focus upon
syntax.

• Any real grammar must be generative.

• The best candidate for the universal grammar is
a transformational grammar.

I suppose it goes almost without saying that any list of revo-
lutionary ideas as long as this is bound to come under attack
from many quarters. And indeed Chomsky has been assailed not
only by his linguistic peers, but also by psychologists, philoso-
phers, and computer scientists, as well as a random assortment
of the other fauna inhabiting the intellectual zoo. Since the main
objective in this chapter is to look at the problem of language
acquisition, we’ll center most of our attention upon the first half

SPEAKING FOR MYSELF

231

LOCALISTS

GLOBALISTS

SUBJECT

MATTER

body of utterances

speakers’ knowledge of
how to produce and
understand sentences;
their linguistic
competence

GOAL

classification of the

individual elements
composing the body of
utterances

specification of the
grammatical rules used
for constructing
sentences

METHODS

discovery procedures

investigation of the
properties of the
universal grammar

TABLE 4.1

of the above list, the problems of mind. In Chomskian terms, our
interest is in the way the child acquires his knowledge of lan-
guage, rather than how he demonstrates his competence. As a re-
sult, most of the opposing positions scouted below will be
objections to one or another of the items on the “psychology”
list. However, since Chomsky has chosen linguistics as the spe-
cific arena in which to defend his theories of mind, we will neces-
sarily touch upon a few linguistic objections to his claims as
well.

At this point the reader may well be wondering whether
Chomsky is the only scholar in the field. Surely there must be
other thinkers of sterling reputation whose work parallels
Chomsky’s but with interesting differences. And indeed there
are. So why haven’t I told you about them? The answer to this
eminently sensible query is quite simple. In virtually no other
area of modern intellectual life that I’m aware of has one man’s
work so completely shaped a field as Chomsky’s views have done
in linguistics. The ideas and programs sketched in this chapter
are but the tip of the iceberg set in motion by his vision, and
there is no better example in the second half of this century of a
true Kuhnian paradigm shift than what has been brought about
in linguistics by Chomsky’s efforts. Consequently, to speak of
linguistic thought from the 1960s onward is really to speak of

232

PARADIGMS LOST

the concepts, ideas, and techniques introduced by Chomsky. As
far as I can tell, all work in mainstream contemporary linguis-
tics involving language acquisition is directed toward either pro-
viding support for or looking for holes in the claims he has
made. So in a very specific sense, Chomsky’s views define what
we mean by much of modern linguistics, in much the same way
that Newton’s ideas defined classical particle mechanics.

Now let’s give the floor over to the Defense to put forth its
array of arguments designed to convince us that Chomsky’s
ideas about language and mind don’t merit much attention
after all.

POSITIVELY REINFORCING

New York’s Greenwich Village has always been a haven for as-
piring artists, writers, and other intellectual hangers-on, at least
until the recent influx of wheeler-dealer yuppies, trendy restau-
rants, and chic boutiques, along with their consequent explosive
effect on the costs of maintaining even an artist’s garret. But in
the Roaring Twenties gentrification had yet to strike the Vil-
lage, and an ambitious young writer from Pennsylvania was en-
couraged by the famed poet Robert Frost to strike out for the
literary life and try swimming with the sharks in what was even
then the highly competitive world of publishing. As in most
cases of this sort, after a couple of years of rejection slips the
romance of living by the pen gave way to the realities of eating
regularly, and the writer-to-be traded in his Village pad for the
comforts of Harvard Yard and the pursuit of a graduate degree
in something useful. The literary world’s loss was the psycholog-
ical community’s gain, as that young man, B. F. Skinner, went
on to found a school of psychology that dominated American
thinking on matters of mind for more than two decades.

In the early 1920s, John B. Watson made the radical sugges-
tion that human behavior does not have mental causes. Stimu-
lated by the ideas of logical positivism, this thesis holds that in
studying behavior all notions of mind, mental states, and mental
representations should be eliminated, and investigations focused
solely upon externally observable stimulus-response behavior pat-
terns. At this time the topic of the day in psychological circles
was the understanding of the learning process, and most psy-

SPEAKING FOR MYSELF

233

chologists were operating under the paradigm established by the
Russian Ivan Pavlov, whose experiments with drooling dogs will
undoubtedly strike a responsive chord (dare I say ring a bell?)
with the reader. In his influential 1925 book Behaviorism., Wat-
son made the infamous claim:

Give me a dozen healthy infants, well-formed, and my own speci-
fied world to bring them up in and I’ll guarantee to take any one
at random and train him to become any kind of specialist I might
select — doctor, lawyer, artist, merchant-chief and, yes, even beg-
gar-man and thief, regardless of his talents, penchants, tenden-
cies, abilities, vocations, and race of his ancestors.

In these few words Watson established the basic elements of be-
havioral psychology: elicitation of any desired behavior solely by
externally applied stimuli and responses, coupled with positive
and negative reinforcements or, more prosaically, rewards and
punishments. It was into this psychological mindset that Skin-
ner dropped when he entered the Harvard Psychology Depart-
ment in 1928.

The behaviorist programs of both Pavlov and Watson as-
serted that learning takes place as a result of environmental
stimuli experienced by the organism, which then responds in
various ways. The responses that either the experimenter or Na-
ture rewards are reinforced, and the maladaptive responses are
soon weeded out by punishments. It was in exactly this fashion
that Watson envisioned training his collection of a dozen infants
to become lawyers, doctors, beggar-men, or thieves. Unlike most
academics, Watson evidently felt confident enough of the sound-
ness of his ideas that shortly after publication of his 1925 book
on behaviorism, he left his professorship at Johns Hopkins for a
life in the business world, trying to transmute the lead of condi-
tioned behavior into the gold of the marketplace. Fittingly
enough, he chose a profession to suit the task, spending the re-
mainder of his years in the advertising business!

Although Skinner is generally regarded as a behaviorist cast
from the mold of the stimulus-response school, he departed in
crucial ways from the program laid down by his predecessors. In
the learning theories of both Watson and Pavlov, the process
unfolds in a fixed sequence: first stimulus, then response. Fol-
lowing the response, the behavior being conditioned is rewarded
while nonconditioned behavior is punished. Skinner objected

234

PARADIGMS LOST

that such a theory of learning could never account for the twin
problems of novelty and purpose in the response.

How did Tolstoy come to write War and Peace f How did John
Lennon and Paul McCartney write those great old Beatles hits?
And how did Bobby Fischer ever find those dazzling combina-
tions on the chessboard? The classical behaviorist has only the
feeble explanation that these tasks can each be broken down into
a series of small behavioral units, all of which exist initially as
unconditioned responses. These individual units are then
brought out as a coherent whole by a mysterious and equally
coherent sequence of individual stimuli.

Such thinking creates a similar problem in explaining a sub-
ject’s purpose. The stimulus-response behaviorist is unable to
talk about the goals and consequences of behavior. The analytic
framework is limited to a discussion of stimuli followed by be-
havior, and there’s just no room for descriptions of the conse-
quences of actions. But it seems highly implausible that even the
most detailed account of a set of stimuli preceding a bicycle ride
around the lake will explain what the rider is doing and why. In
the final analysis, classical behavioral psychology fails to offer a
framework for understanding human actions that can peacefully
coexist with a belief in the psychological reality of innovative
and goal-oriented behavior.

In Skinner’s radical behaviorism, he attempts to address these
difficulties by eliminating the notion of antecedent stimuli, re-
placing it with the idea of operant behavior, i.e., behavior ac-
quired, shaped, and maintained by stimuli occurring after the
responses rather than before. At the same time, he argues that
only positive reinforcement produces behavior that leads to a
more satisfying life, dropping the kind of Clockwork Orange-
style aversion therapy inherent in the work of Pavlov and Wat-
son. Thus, for Skinner desirable behaviors are rewarded after
they are performed, and the reward is what acts to enhance the
likelihood of that behavior’s being repeated.

It’s easy to see that Skinner’s idea of operant behavior is the
psychological analogue at the individual level of biological evolu-
tion at the level of the species. In Skinner’s scheme, “good” be-
havior is reinforced, just as in Nature, “good” mutations are
selected. But in both cases, there is no reinforcement until after
the action has taken place. Operant conditioning is designed to

SPEAKING FOR MYSELF

235

explain the emergence of novel behavioral patterns in the indi-
vidual in the same way that natural selection explains the emer-
gence of new traits in a species. In both cases, the role of the
environment is more to select than either to reward or punish,
although Skinnerian reinforcement can be likened to a reward
since it encourages continuation of certain types of behavior
just as natural selection encourages certain types of mutations.
Note, however, that despite his major departure from the learn-
ing schemes advocated by Pavlov and Watson, Skinner still re-
tains the keystone in the behaviorist arch: the inadmissibility of
any notion of a nonphysical mind, mental states, or mental enti-
ties in the scientific explanation of behavior.

Skinner, of course, is well known for the many ingenious ex-
periments he set up to try to validate his behaviorist theories.
For instance, during the Second World War he devised a kind
of pigeon guidance system for targeting bombs. This unlikely
system involved placing trained pigeons inside the warhead with
a map depicting the terrain to be bombed. The pigeons would
peck at the map in all the right places to activate a steering
mechanism, with the right sequence of peckings presumably en-
suring that the bomb would be brought back on target if it
started to veer off course. Another one of his more widely re-
ported schemes was the so-called air crib, a variation of his fa-
mous Skinner box used to train the pigeons. In the crib, a kind
of glass-enclosed box, the interior was carefully regulated to be
at just the right temperature and humidity to form an ideal at-
mosphere for an infant. Furthermore, germ filters cleaned the
air, doing away with the need for blankets, clothing, and fre-
quent baths. The crib also contained a variety of equipment to
keep infants amused and well exercised. Skinner tested the de-
vice by placing his own daughters in it, which at the time (1945)
caused quite a public flap. Contrary to sensationalist reports, the
girls, who now both lead rather normal, well-adjusted lives, did
not become suicidal or psychotic, and both report that they think
the experience was beneficial.

Given his predilections for seeing operant behavior in every
corner of life, it should come as no surprise that Skinner has
devoted a considerable amount of his impressive reservoirs of
intellectual and polemical energy to the problem of language
learning. He is particularly interested in this question since he
holds the view that language and self-knowledge are intimately

236

PARADIGMS LOST

intertwined. In the Skinnerian view, all words are acquired on
the basis of the “law of effect,” i.e., by rewarding, ignoring, or
correcting the performance of novices by more mature users of
the language. As a result of the way the human brain is struc-
tured to learn, a child comes to identify its pet dog with a word
such as “Spot,” with this identification taking place through a
sequence of positive reinforcements from parents and older
friends who have developed a more mature use of the language.
So in the Skinnerian version of language acquisition, language
is learned in exactly the same way (operant conditioning) and
with exactly the same psychological mechanisms (unspecified) as
the child learns any other skill, such as bicycle riding, tying
shoelaces, or telling time.

The behaviorist position of Skinner raises the question about
how the child learns words for private events. Such events can-
not be reinforced by external means like pointing to an object or
showing a picture in a book, yet must somehow come to be un-
derstood by the child in the same sense as they’re understood by
the rest of the speech community. Skinner’s answer to this di-
lemma is to skirt the issue by asserting that teaching words for
private events is akin to trying to teach color words in a world
of partially and unpredictably color-blind people. His claim is
that our confidence in the reliability of our inferences about pri-
vate events is based upon observable behavior, and we simply
cannot be sure that people use the language of private events to
mean the same things.

The radical behaviorist view of language acquisition was put
on record in Skinner’s 1957 book Verbal Behavior. Noam
Chomsky’s scathing review of this book in the prestigious jour-
nal Language in 1959 gave Chomsky his first widespread recogni-
tion as an opponent to the empiricist claims of most scientists of
that era. With considerable relish, Chomsky argued that the be-
haviorist conception of language acquisition cannot possibly be
correct, and that “with a literal reading . . . the book covers
almost no aspects of linguistic behavior, and that with a meta-
phoric reading, it is no more scientific than the traditional ap-
proaches to this subject matter, and rarely as clear and careful.”
Later Chomsky expanded his attack on Skinner’s ideas by stat-
ing that

Skinner’s approach has led absolutely nowhere. ... It has yielded

no theoretical knowledge, no nontrivial principles as far as I am

SPEAKING FOR MYSELF

237

aware — thus far, at any rate. . . . Skinnerian behaviorism is off
the wall. It’s as hopeless a project as trying to explain that the
onset of puberty results from social training.

The essence of Chomsky’s critique is that the learning process as
Skinner describes it is at crucial points left to vague notions like
“analogy” and “generalization,” notions that are inherently in-
capable of offering any sort of explanatory power.

Skinner never responded to this savage review of his life’s re-
search program, although to this day he maintains that psycho-
therapists and psychologists rely too much on inferences they
make about what is going on inside their patients’ heads, and too
little on what the patients are actually doing. He further con-
tends that “I think cognitive psychology is a great hoax and a
fraud, and that goes for brain science, too.” Nevertheless, most
researchers tend to feel that Chomsky’s review of Verbal Behav-
ior sent behavioral psychology into a tailspin that it may never
pull out of. The reason is one that Skinner himself would surely
approve: Behaviorism just isn’t very reinforcing nowadays. So
with the fading of Skinnerian visions from the psychological
stage, let’s move across the ocean to the land of the cuckoo clock
and chocolates to hear from our next witness against Chomsky.

OUT OF THE MOUTHS OF BABES

In 1918 a small Lausanne publishing house brought out the
novel Recherche, an account of the conflicts felt by a young
Catholic over the relationship between science and religion. Com-
mercially the book sank like a stone, a fact that the young au-
thor later probably had little cause to regret. However, leaving
its dubious literary merits aside, Recherche deserved a better
fate if for no other reason than that its detailing of the relation-
ship between the part and the whole in organic life represented
the initial public glimpse of the thoughts that later led its au-
thor, Jean Piaget, to play a founder’s role in the development of
what is now termed cognitive psychology. So as with Skinner, a
small loss to the world of letters became a giant gain to the
world of science and the study of the mind.

As a youth growing up in the Swiss town of Neuchatel, Piaget
was a passionate collector of shells, fossils, and other such ef-
fluvia of nature, an interest that led to an unofficial position as

238

PARADIGMS LOST

assistant curator of the Neuchatel natural history museum at
the precocious age of ten. As a consequence of his childhood bio-
logical obsessions, the young Piaget developed a lifelong focus
on the structure of organisms, as well as a deep attraction to the
philosophy of Henri Bergson and its attempt to fit questions of
mind, matter, science, and soul into an overall, integrated world
view. During this period of youthful contemplation, Piaget had
already begun formulating the notion that all organisms consist
of parts related to the whole, and that all knowledge derives
from the assimilation of external experiences into the organism’s
structure. Piaget’s key idea involved a comparison between men-
tal processes and the body. Just as the body requires balance
and self -regulation in all of its biological functions, so too does
mind require equilibration of its intellectual levels. Hence by the
time he was out of his teens, Piaget’s lifetime research direction
was already set: to explore the interrelationship between biology
and logic, with the wordings of the human mind as the bridge
linking the two.

Upon completion of his studies, the specific vehicle with which
Piaget was to pursue his grand research plan emerged from a
job he was offered as an assistant to an assistant to Alfred
Binet, the developer of the IQ test. Piaget was hired to stan-
dardize some of the tests, and during the course of his work he
noted that the kinds of mistakes that children made on the tests
were not random but tended to fall into definite categories, de-
pending upon the age of the child. Rather than dismissing this
observation as a statistical irregularity, Piaget conjectured that
it was a sign that qualitatively different structures of intellect
were present at different stages of the child’s cognitive develop-
ment. The pursuit of this theme was to occupy Piaget for the
rest of his life.

The pivot around which all of Piaget’s ideas revolve is his vi-
sion of the mind as not just a passive device for handling sen-
sory inputs, but a mechanism that actively transforms the
inputs it receives by performing exploratory operations upon
them. Thus, Piaget thought of human intelligence as a process
of reality construction rather than as a passive receiver and
processor of information from the outside world. A core ingredi-
ent supporting this concept of the active, exploring mind is the
idea of an internal mental representation. Piaget, unlike Skin-
ner, felt that to postulate such unobservable, even hypothetical,

SPEAKING FOR MYSELF

239

internal mental states was a necessary step on the path toward
providing explanations for mental development. Furthermore,
he felt the introduction of such entities into psychology was no
more an obstacle to making the study of mind “scientific than
the introduction of concepts like neutrinos and electrons was a
barrier to making physics “scientific.” In fact, if we wanted to
pinpoint the precise moment when the “cognitive revolution” in
psychology began, we can probably do no better than to mark
the day when Piaget pushed forward his claim for mental repre-
sentations as valid objects of study in the creation of a science of
human thought.

Shortly after completing his work in Paris, Piaget was offered
the position of director of the Rousseau Center for “genetic psy-
chology” in Geneva, where he spent the remainder of his long
and fruitful career. Upon arrival in Geneva he quickly inaugu-
rated a program of research on the intellectual development of
children, using many ingenious experiments to identify the vari-
ous stages in his theory. According to Piaget, the child goes
through at least four main stages in mental evolution from a
little savage to a more or less right-thinking adult. These
qualitatively distinct stages are:

• Sensorimotor stage: birth to two years. This is the period when
infants construct the concepts of an object, space, and causal-
ity. This involves an increasingly coordinated linkage between
perception and action. For example, the child’s perception of
objects like a doll or a rattle becomes synonymous with the
actions that can be performed upon them, such as shaking the
rattle or holding the doll.

• Preoperational stage: two to five years. At this tune the child’s
thought processes begin to use symbols in the form of mental
images arising from imitation or words. During this period
reasoning from memory and analogy also begins to occur, as
does the development of language skills.

• Operational stage: five to ten years. In this stage the child per-
forms mental operations on objects that are physically pre-
sent. Classification of hierarchical structures occurs, as does
the understanding of ordinal relations. Near the end of this
period, the concept of conservation of continuous properties
like weight, quantity, and volume emerges, so that the child

240

PARADIGMS LOST

begins to recognize that there is not less liquid present when
water is poured from a tall, narrow tube into a short, flat
bowl.

• Formal operations stage: ten to fourteen years. At this time the
real world is conceived of as a subset of possible worlds.
Propositional thinking, with assertions and statements that
can be true or false, becomes possible, and there is a better
grasp of the fact that appearances can be deceiving.

Just as the acquisition of language involves progression through
a set of strictly ordered stages, Piaget held that general intellec-
tual development follows the path outlined above, and that, like
the Stations of the Cross, each of the steps must be passed
through with no omissions or change of order.

As to exactly how experience was processed within each stage
to generate knowledge, Piaget advocated a two-pronged theory
in which the child pits the antithetical processes of assimilation
and accommodation against each other in a kind of dynamic arm
wrestling. Assimilation, for the child, involves trying to fit novel
aspects of reality into old behavioral and cognitive schemes
rather than changing them. On the other hand, accommodation
requires changing an existing mental or behavioral pattern to
adapt it to the specific characteristics of new objects and new
relationships, in this way taking account of novel aspects of re-
ality. The tension between these two approaches to dealing with
novelty in the environment is then resolved by what Piaget
called equilibration, a type of dynamic steady state that balances
out the competing forces. This again calls to mind Piaget’s ini-
tial preoccupations with the process of autoregulation in biologi-
cal systems. Now let’s try to put these rather general ideas about
learning and mental development into the specific context of lan-
guage acquisition.

In the epistemology of Piaget the child does not come “hard-
wired” to understand concepts, but has to create them as in his
construction of the ideas of space, time, conservation, and so on.
In this framework, the environment provides feedback about the
quality of the mental structures the child creates; it does not
simply imprint the right structures on the mind. Thus, for Pia-
get the world is not just “out there” waiting to impress itself on
a blank slate. Intellectual development is a constant interplay
between the child and his environment, with the child playing an
active, structuring role. Moreover, the Piagetian sees all areas of

SPEAKING FOR MYSELF

241

mental development as being closely interconnected with each
other. So as far as language acquisition goes, Piaget sees it as all
of a piece with the other stages of intellectual growth, and he
places no particular emphasis on language as opposed to the
other skills the child learns. As a result, the Piaget school con-
tends that the mind develops more as a whole across a spectrum
of intellectual tasks than as a modular structure.

Since the Piaget position differs from both Skinnerian behav-
iorism and Chomskian rationalism in a number of interesting
ways, let’s briefly summarize the differences. The major points
of each position are shown in Table 4.2.

Since the differences telegraphically noted in the table are so
central to the entire issue of cognition, we’ll defer detailed dis-
cussion of them to a later section where the problems and possi-
ble rapprochements can be given the attention they properly
deserve. But first let’s have a short intermezzo and look at some
of the complaints that have been registered in the linguistics
community against the syntax-dominated position of Chomsky.

IT’S ALL A QUESTION OF SEMANTICS

One of the most surprising outcomes of Piaget’s research pro-
gram was the discovery that the earliest function of speech is
not communication, but symbolization. Thus, the first entities
that the child perceives and that stand for a certain content or
meaning are private symbols. These lead to internalization and
representation of thoughts, with social communication arising
only at a later stage. So in this view language is more of a tech-
nique or strategy for structuring thought than a vehicle for
communication. This discovery is entirely consistent with
Chomsky’s idea of a universal grammar and his focus on syntax
as the real core of language. But the deemphasis of content at
the expense of form has not always occupied a favored position
in the linguist’s order of things and, in fact, has had to be toned
down somewhat from Chomsky’s original proposals. To see how
things stand today, let’s go back for a moment to the days when
meaning was still king in the world of linguistics.

Benjamin Whorf was a chemical engineer by training, a lin-
guist by avocation. While spending his entire professional life as
a fire inspector for a large Hartford insurance company, prior

242

PARADIGMS LOST

MENTAL STATES?

LANGUAGE ORGAN?

ENVIRONMENT/

HEREDITY?

Chomsky

yes

heredity

Skinner

environment

Piaget

yes

both

TABLE 4.2. Positions on mind and language

to his untimely death at the age of forty-four, Whorf served as a
striking example of the gifted amateur competing on equal terms
with the professionals by devoting his spare time and en-
ergy to a detailed study of the languages of the American Indi-
ans, particularly the Hopi of the American Southwest. In these
efforts, Whorf was following in the wake of his teacher Edward
Sapir, an anthropologically oriented American linguist of the
pre-Chomsky era, who advocated the position that one’s view of
the world is strongly shaped, if not totally created, by language.
This claim calls to mind the contention of the later Wittgenstein
in his statement that “the limits of my language mean the limits
of my world.” This argument was expanded upon by Sapir when
he stated:

. . . the “real-world” is to a large extent unconsciously built up on
the language habits of the group. No two languages are ever suf-
ficiently similar to be considered as representing the same social
reality. The worlds in which different societies live are distinct
worlds, not merely the same world with different labels attached.

The linguistic ideas of Sapir and Whorf have come to be en-
shrined in what is now termed the Sapir-Whorf Hypothesis,
consisting of two main assertions relating language to thought:

SAPIR-WHORF HYPOTHESIS

• Linguistic determinism: Language determines the way we think.

• Linguistic relativism: The distinctions encoded into one lan-
guage are not found in any other language.

The famous example of the Eskimo language, which has sepa-
rate words for falling snow, snow on the ground, snowed packed
hard like ice, slush, and so forth, illustrates the point.

The fact that translations from one language to another can
be made, as well as the fact that the conceptual uniqueness of a

SPEAKING FOR MYSELF

243

language like Eskimo can still be explained using another lan-
guage like English, makes it unlikely that a strong form of the
Sapir-Whorf Hypothesis is correct. While it’s undeniable that
there are conceptual differences between languages due to cul-
tural and environmental factors, this does not necessarily imply
that the differences are so great that mutual comprehension is
impossible. It’s always possible to use various types of circumlo-
cutions to say in many words in one language what can be said
more compactly in another. As an example, consider the diagram
in Figure 4.4 displaying the ways of saying “He invites people
to a feast” in English and Nootka, an Indian language of the
Pacific Northwest. Nootka is able to express in a single word an
idea that in English requires a far more elaborate construction.

Even though a strong form of the Sapir-Whorf Hypothesis
seems unlikely to be true, a weaker form is probably valid, as-
serting that language does affect the way we perceive and re-
member, as well as facilitate the performance of mental tasks. If
so, the weak form of the Sapir-Whorf Hypothesis might lead us
to speculate that there’s a lot more feasting going on among the
Nootka than the British, in light of the relative ease of issuing
the requisite invitation in Nootka.

With these kinds of ideas, the feet of both Sapir and Whorf
are firmly planted in the localist school of linguistics with its
emphasis on the differences in languages as being of paramount
importance. And these differences center upon matters of mean-
ing, i.e., semantics. This, of course, is exactly the situation
whenever we study a literary text, as by definition the reader
and student of literature “work at the surface,” as noted by the
literary critic and language scholar George Steiner. Such texts
deal with phonetic and semantic facts, the words and sentences
that we can actually see and hear. That is the only reality availa-
ble to us, so on the surface we are all ultra-Whorfians. The
transformational grammarians assure us that the surface pres-
ence of the text is merely an external product emerging out of
deeper structures, and that to understand language it’s neces-
sary to descend to these primal levels. In short, Chomsky’s not
so tacit assumption that syntax can be studied profitably de-
tached from semantics comes under a dark cloud. Let’s briefly
consider a couple of the more interesting objections and the re-
sponses to them.

* * *

244

PARADIGMS LOST

THE ENGLISH LANGUAGE

BOIL — ED — EAT — ERS — GO-FOR — HE DOES
TL'IMSH — YA 'IS ITA — TTL MA

IN PHONETIC WRITING : Vimsya-'isita-'iXma

FIGURE 4.4. An invitation in English and Nootka

Recall that Chomsky’s so-called Extended General Theory
comprises the sequence shown in Figure 4.5, where we see the
surface structure emerging out of the deep structure, each being
processed by both the phonetic and semantic rules to generate
what we think of as everyday speech. This picture clearly shows
the role of both phonology and semantics as logically following
syntax in Chomsky’s world.

The most obvious line of attack against this picture is to argue
that there is no clear-cut distinction between syntactic and se-
mantic rules; hence, the level of syntactic deep structure cannot
be defended. This is the position of the so-called generative seman-
tics ts, who have tried (rather unsuccessfully) to set up syntac-
tic-semantic rules that take semantic representations as their
input and yield surface structures as their output, using no in-
tervening level of deep structure. A closely related idea is that
pursued by the interpretive semanticists, who argue for moving
more and more of the syntactic rules into the semantic compo-
nent, thereby moving the deep structure closer to the surface
structure of the language. Let’s see how one variation on this
basic theme offers the promise of patching up at least a few of
the holes in the Chomskian facade.

Part of the problem with the Extended General Theory was a
1971 mathematical result produced by Peters and Ritchie show-
ing that the original transformational grammars are just too

SPEAKING FOR MYSELF

245

(

Semantic

Representation

FIGURE 4.5. Chomsky’s Extended General Theory

general. This theorem demonstrated that any language whose
sentences could be listed mechanically could be generated by
some Chomskian grammar. Thus, Chomsky’s claim that natural
languages had transformational grammars essentially amounted
only to the claim that they could be characterized mathemati-
cally. A major part of the difficulty is that the Chomskian gram-
mars do not necessarily provide a mechanical decision procedure
for the grammaticality of the sentences of the language because
they are so general. What this means is that while sentences can
be mechanically generated by such a grammar, the grammatical-
ity of any sentence given in advance cannot be decided by apply-
ing the rules of the grammar. Or at least, grammaticality cannot
be decided by any procedure guaranteed to terminate in a finite
number of steps. As a consequence of results of this type, inter-
est in transformational grammars waned in the 1970s, only to be

246

PARADIGMS LOST

reborn with the work of Richard Montague, who showed that it
was possible to associate an equally explicit semantic theory with
the syntax. In his words, “There is in my opinion no important
theoretical difference between natural languages and the artifi-
cial languages of logicians; indeed, I consider it possible to com-
prehend the syntax and semantics of both kinds of languages
within a single natural and mathematically precise theory.” This
manifesto, together with the theoretical framework supporting
it, has pumped new life into the area of generative grammars,
but now with syntax and semantics coexisting on a more equal
footing. Since this is neither the time nor the place for an ac-
count of the highly mathematical details of Montague’s work,
let’s instead look briefly at another competitor to Chomskian vi-
sions, a viewpoint whose ideas are in harmony with Montague’s
and which combines features of both Chomsky’s and Piaget’s po-
sitions but without totally supporting either.

Geoffrey Sampson is a British linguist who has challenged the
notion that there are as many linguistic universals as Chomsky
claims. However, Sampson does agree with Chomsky on the exis-
tence of at least one such universal, the hierarchical nature of all
languages. And he proposes a theory of language acquisition
that he feels accounts for this universal feature without having
to invoke the specialized language organ that Chomsky so
cherishes.

Sampson’s argument is based upon a parable first introduced
by Herbert Simon to explain why all complex systems generally
seem to display a hierarchical structure. Simon considered the
assembly of a watch consisting of ten subassemblies, each of
which consists of ten individual components. Assuming that the
watchmaker is periodically interrupted in his task of assembling
the watch, with each such interruption necessitating his starting
from scratch on the construction of the part of the watch he is at
that moment putting together, Simon shows convincingly that
with even a minuscule chance of interruption the watchmaker
will never finish assembling the watch if it’s regarded as a single
object of one hundred pieces. On the other hand, the chances of
finishing are excellent, even with many interruptions, if the
watch is divided hierarchically into subassemblies and the
watchmaker has only to put together the subassemblies to make
the final product. This so-called Watchmaker Parable forms the

SPEAKING FOR MYSELF

247

heart of what Sampson claims is a major improvement upon
Chomsky’s ideas.

Applying the Watchmaker Parable to syntactic structures,
Sampson argues that the communication system of our ancestors
presumably consisted of words and short sentences, and that
language users occasionally hit on new combinations of phrases
to produce slightly longer sentences than had earlier been the
rule. There are then two ways the new sentences could become
entrenched in the language: (1) a new sentence might effect the
transfer of enough useful information sufficiently often that
there was some selective advantage for the organism to transmit
just that sort of information, or (2) a new sentence might put
simpler grammatical elements together in new and more complex
ways resulting in a linguistic innovation — i.e., it would repre-
sent a new semantic category not present in any of its parts
taken individually. In this way, Sampson argues, a child can ac-
quire language in just the same manner as the watchmaker puts
together the watch — by composing subassemblies from individ-
ual components, and then putting together the subassemblies. In
experiments by Berlin and Kay involving the way color words
like “red” and “yellow” are learned, it was found that learning
followed an evolutionary sequence in all languages regardless of
their grammars, a strong point in favor of Sampson’s theory.

This idea is completely consistent with the way a complex ex-
pression is generated using a Montague grammar. In such a
grammar, we begin with the lexical items and “assemble” the
lower-level structures. Within these new structures, it’s possible
to discern the earlier items whose syntactic combination in ac-
cordance with Montague’s rules involves only minor peripheral
modifications. The new lower-level structures are then them-
selves syntactically combined, again with peripheral modifica-
tions, into yet higher-level ones, and so on. Finally, the sentence
itself is produced. An example of the kind of tree that comes out
of a Montague grammar is shown in Figure 4.6 for the sentence
“Every man loves a woman such that she loves him.”

Substitution of a Montague grammar for a Chomskian one,
together with Sampson’s account of the origin of hierarchical
structure in languages via the Watchmaker argument, leads to a
theory of language acquisition that can dispense with the innate
language organ. Instead we require only the kind of general
problem-solving capability promoted by Piaget, with the child

248

PARADIGMS LOST

every man loves a woman such that she loves him, 10, 0

every man, 0 he„ loves a woman such that she loves him„, 4

man he0 love a woman such that she loves him0, 5

love

a woman such that she loves him0, 2

woman such that she loves him0, 3, 1

woman

he, loves him0, 4

love him0, 5

UVc IiIIIIq,

love he0

FIGURE 4.6. A Montague tree

subconsciously and implicitly testing various hypotheses about
grammaticality against the actual linguistic data encountered.
The assumption that ensures that the child will home in on the
right hierarchically structured language is the presupposition
that the child’s program of exposure and hypothesis testing will
take place in a linguistic environment where the local language
has precisely this “right” structure. Ergo, by following a Pop-
perian program of conjectures and refutations, the child arrives
at the right set of rules.

Before closing this discussion of semantics and Montague
grammars, let’s note one further point: The Montague grammars
are capable of characterizing at most context-sensitive languages,
for which it can be shown that a mechanical decision procedure
does exist for establishing the grammaticality of a given sen-
tence. Recent work by Gerald Gazdar and others indicates that
natural languages like English are a subclass of these, termed
context-free languages. What this means is not that the sentences
have a meaning independent of the context where they are used,
but that the phrase structure rules are formulated in such a way
that a category label can be rewritten without regard to the sur-
rounding context of words in the sentence. In other words, the
way the phrase structure tree branches depends only on what

SPEAKING FOR MYSELF

249

the situation is at the branch point, and not on what lies on
other branches of the tree. Thus, because Montague grammars
are decidable (i.e., they possess a decision procedure for gram-
maticality), contain only phrase structure rules and no transfor-
mations, treat syntax and semantics on the same footing, and
are harmonious with an evolutionary view of linguistic develop-
ment, they can be taken seriously as an alternative to the Ex-
tended General Theory of Chomsky.

Following this brief appearance on the stand by Chomsky’s
linguistic opponents, let’s get back to the philosophers and psy-
chologists and examine a few of the arguments put forth against
his theory of mind.

SHOOT-OUT AT THE ROYAUMONT CORRAL

A few miles down the road to Mexico from Tucson, Arizona, lies
the ghost town of Tombstone where in 1881 at the famous O.K.
Corral the Clantons and the Earps, with a little help from Doc
Holliday, fought the most famous gun battle of the Old West,
earning Tombstone the sobriquet “The Town Too Tough to Die.”
To this day, twice a month members of a local community group
strap on their six-shooters and reenact this famous gun battle
for the benefit of tourists like myself, who yearn to feel even for
a moment a bit of the excitement and lawlessness of those leg-
endary times. During a recent visit to this living monument to
the past, I was quenching my thirst with a beer at the Bird Cage
Saloon following the festivities at the corral, when the thought
struck me that perhaps the conflict resolution methods of rip-
roaring Tombstone and those of the modern-day intelligentsia
have a lot more in common than most of us realize. Other than
the admittedly nontrivial difference that the academic and intel-
lectual losers don’t expire from lead poisoning, the similarities
are striking: diametrically opposed forces clashing in a public
arena (scholarly journal articles and open lectures), hotshot
young challengers looking to make a name for themselves by out-
drawing the top gun, even a version of the Boot Hill Cemetery
for those whose ideas bite the dust at high noon on Main Street
(professorships in academic “Siberia”). Looked at in these
terms, one of the most talked-about and intellectually violent
showdowns in recent times took place not on the dusty streets of

250

PARADIGMS LOST

an Arizona boomtown, but in the august halls of a luxurious
French chateau when in October 1975 Noam Chomsky rode in
from Cambridge, Massachusetts, to do battle face to face with
Jean Piaget.

The Centre Royaumont pour une Science de l’Homme is
located just outside Paris in an elegant chateau of the type that
would make any royalist’s heart flutter. At the time of the de-
bate, the now-booming subject of cognitive science was only be-
ginning to emerge out of its parent disciplines, and the center
found the biological ideas of Chomsky and the cognitive perspec-
tive of Piaget to be central to a proper understanding of the
mind and its workings. As a result of the enthusiastic encour-
agement of its president, the famed biologist Jacques Monod, the
center’s staff arranged for a constellation of biologists, computer
scientists, psychologists, and philosophers to bear witness to the
struggle of the two titans, as well as to provide dissenting views
from the chorus. The mainstream of popular psychological
thought prior to the Royaumont gathering centered upon three
principal themes: psychoanalysis, behaviorism, and classical
learning theory. Significantly, at Royaumont not one of these
traditional areas of mind was represented, leading more than
one observer to date the coming of age of cognitive science to
this unique conclave.

The real issue before the house at Royaumont was the inter-
play between the question of the nature of various vehicles of
knowledge, such as images, signs, and schemata, and the problem
of whether knowledge is inborn as Chomsky claims, or con-
structed through interaction between certain inborn modes of in-
formation processing and the actual characteristics of the
physical world as asserted by Piaget. Of course the positions are
not so clear-cut as this, as indicated by a remark of Monod’s:
“In asking myself the question, ‘what makes man man?’ it is
clear that it is partially his genome and partially his culture.
But what are the genetic limits of his culture? What is its gen-
etic component?”

While struggling with this eternal question, the participants
at the debate focused their arguments on three main topics: child
versus adult thought, the nature of mental representations, and
the generality of thought and thought processes. On the first
topic, the Piagetians argued for the stages of mental develop-
ment noted earlier. By way of response, MIT philosopher Jerry

SPEAKING FOR MYSELF

251

Fodor pointed out that it’s logically impossible to generate more
powerful forms of thought from less powerful ones, and that all
forms of reasoning that a person will ever be capable of are al-
ready present at birth and gradually emerge through a process
of “growing up.” Thus, his position strongly supported
Chomsky’s nativistic views of mental development.

As to the issue of mental representations and thought, while
both sides accepted the validity of postulating unobservable, yet
no less real, mental representations as a means to explain mental
processes, there was considerable disagreement as to the nature
and specific role of these representations. For instance, Piaget
asserted that the ability to represent knowledge to ourselves is a
process of construction, taking place over a long series of in-
teractions with the environment, and cannot be really initiated
until the end of the sensorimotor stage at about the age of two
years. But if this were true, the Chomskians argued, we would
expect paraplegics to have a distorted path of language develop-
ment, a prediction that is not borne out by the evidence.
Chomsky also doubted the validity of grouping together a fam-
ily of such representations, arguing for a modular view of a
mind composed of individual “compartments,” each emerging in
its own time to carry out its preassigned mental tasks.

In the Chomskian view, the human language capacity is just
one of these mental modules, and is for the most part divorced
from other forms of thinking. Here Chomsky was claiming that
thought is a collection of heterogeneous “actors” loosely con-
trolled by some central organizing agent. Perhaps ironically, in
this sense his vision of the mind is reminiscent of the organiza-
tion of Piaget’s homeland, the Swiss Confederation, with its col-
lection of individual cantons loosely held together by the central
government in Bern. Piaget, of course, took the opposite tack,
insisting that thought is a broad set of capacities, with identical
mental operations underlying the individual’s encounters with a
wide range of environmental stimuli, such interactions eventu-
ally shaping the homogeneous mind into more specialized compo-
nents. In rebuttal, Chomsky challenged the Piagetians to
address the problem of the poverty of the stimulus, and to ex-
plain how generalized learning strategies could ever overcome
this major hurdle.

For the most part, the biologists at Royaumont tended to
favor Chomsky, perhaps on account of some rather strange anti-

252

PARADIGMS LOST

Darwinian views expressed by Piaget concerning a kind of La-
marckian “transfer of structure” from the environment to the
organism. The social scientists present appeared to be equally
divided between the two competing schools of thought. It should
not go unmentioned that Chomsky’s unparalleled skills as a
debater may also have played a nontrivial role in tilting the
Royaumont scales in his favor. Having honed these skills to a
razor-sharp edge through numerous encounters with the bar-
racudas of the American political and academic intelligentsia,
Chomsky was well prepared to counter the low-key, gentlemanly,
almost apologetic style prevalent in European intellectual de-
bate.

As an interesting aside, it’s perhaps worth noting here the re-
lationship between Chomsky’s strongly biologically oriented po-
sition on mental development and the position of sociobiologists
like Edward O. Wilson on the role of the genes in determining
human behavior patterns. On the basis of surface arguments,
one might well speculate that Chomsky would be most sympa-
thetic to the sociobiologists since, after all, one of his central
claims is that our language capacity is inherently limited by our
genetic endowment. Perhaps surprisingly, in actual fact Chom-
sky appears to be at best lukewarm toward Wilson’s arguments.
While firmly committed to the position that a good deal of our
personal and social behavior is a reflection of our genetic pro-
gram, Chomsky has gone on record with the statement that “I
don’t think Wilson understood what he was talking about in
that final chapter.” Here he was referring to the last chapter of
Wilson’s book Sociobiology and its treatment of human behavior.
Somehow this statement seems strangely at odds with Chomsky’s
later elevation of heredity over environment, as noted, for exam-
ple, in his Managua Lectures, where he claims that “the evidence
seems compelling, indeed overwhelming, that fundamental as-
pects of our mental and social life, including language, are de-
termined as part of our biological endowment, not acquired by
learning, still less by training. . . .”

Following this pronouncement, which, on the surface at least,
certainly appears to be consistent with many of the strongest
claims of the sociobiologists, Chomsky goes on to speculate as to
why so many intellectuals find such assertions so difficult to
swallow. His conjecture is that intellectual libertarians have

SPEAKING FOR MYSELF

253

become ideological and social managers, seeking to serve or as-
sume power for themselves by taking control of popular move-
ments. For such people committed to control and manipulation,
Chomsky claims, it’s very useful to think that humans have no
intrinsic (i.e., innate) moral and intellectual nature, and that
they are simply objects to be shaped for their own good. To my
untutored eye, this looks to be about as strong a claim for at
least the spirit, if not the program, of sociobiology as could be
offered. However, to pursue this line of argument would take us
too far afield at this point, so let’s return to our main concerns.

In summary, it’s clear that the “shoot-out” was more of an
exploration than a definitive resolution or rapprochement. But
as in all good gunfights, both Piaget and Chomsky stuck firmly
to the styles that had got them to the top, and who would have
ever thought otherwise? But what about the jury of peers?
When the intellectual pyrotechnics and academic smokescreens
cleared away, did either of the combatants live to fight another
day? Again, as in all the movie westerns at least, only one man
rode off into the sunset and that lone gunslinger was Noam
Chomsky. But regardless of how one judges the result of this
particular encounter, what is clear is that at least a few of the
pillars upon which the cognitive sciences now rest were firmly
erected as a result of the debate. Since it’s important for our
later deliberations, let’s take a few pages now to discuss this
“cognitive skeleton” before summing up the overall issue of lan-
guage and our verdict as to how it’s acquired.

RULES AND REPRESENTATIONS

In the search for the seat of grammar, the one thing all the dis-
putants seem to agree upon is the nature of the grammars them-
selves: They are a set of rules enabling us to distinguish a
sentence that is acceptable in a given language from a sentence
that’s not. For instance, a trivial and useless grammar for En-
glish might contain the rule: A sentence is OK if it contains an
even number of words, unacceptable otherwise. The major goal
of linguistic researchers is to identify more extensive collections
of such rules that, taken together, determine what goes and what
doesn’t for utterances in the target language. But from the
higher-level standpoint of general thought processes, the case of

254

PARADIGMS LOST

linguistics raises the broader issue of the degree to which all
human thought processes are governed by rules. If we grant the
cognitivists their use of mental representations, is it true that
every thought you think and action you take involves these rep-
resentations’ being shoved around inside your skull according to
the dictates of a collection of rules? Since I intend to devote the
next chapter to an extensive discussion of this very question, for
now I’ll sketch only one or two aspects relevant to our linguistic
concerns of the moment.

As we’ll discuss more fully in the next chapter, to contend
that the mind operates according to rules means that we can
view the mind as an information-processing machine of the sort
depicted in Figure 4.7. Here the inputs to the system, or “ma-
chine M ,” represent the environmental stimuli processed by M
to produce the observed outputs (actions or behaviors). The
inner workings of M are fenced off from the inputs and outputs
by the dotted lines to indicate that, generally speaking, an inves-
tigator has direct access only to the inputs and outputs, not to
the internal mechanisms of M. The workings of M should be
thought of as unfolding in one of two ways:

• External description — processing of the inputs directly into
outputs by a set of external behavioral rules.

• Internal description — processing of the inputs into outputs by
the following steps: (1) Inputs are applied to M from the envi-
ronment. (2) The inputs are “encoded” by internal rules
within M as mental representations. These, in turn, are
manipulated within M by other rules to form new mental rep-
resentations. (3) The new representations are then “decoded”
by additional internal rules of M to produce the externally
observed behavior of the system.

It’s crucial to note that there are two conceptually quite, differ-
ent sets of rules operating here. There are the external rules di-
rectly relating inputs and outputs. Such rules can be thought of
as the sort of stimulus-response patterns so loved by behavior-
ists. On the other hand, there are the internal rules living inside
the system. These rules are the ones that the cognitivists crow
about when extolling the virtues of manipulating mental repre-
sentations as explanatory objects for the mind. The sixty-four-
dollar question then becomes: Do these two types of rules have
anything to do with each other, and if they do, does this machine

SPEAKING FOR MYSELF

255

FIGURE 4.7. Schematic diagram of an information-processing machine

metaphor serve as an adequate model for the way the human
mind actually works?

In support of the machine metaphor as a model of the mind,
let’s briefly see how every one of the positions taken in this chap-
ter on matters of language and the workings of the mind can
comfortably be interpreted within the confines of the structure
shown in Figure 4.7. First of all, every behaviorist from Watson
to Skinner has argued that whatever may be in the box labeled
“internal description,” it has no place in any scientific theory of
behavior, and that such a theory must be based solely upon the
“external description” of the system. Cognitivists like Piaget
say that it’s perfectly acceptable to invoke theoretical objects
like the rules and representations composing the “internal de-
scription,” but that those rules and representations can be cre-
ated only by the system’s interaction with its environment.
Finally, the Chomskians argue that not only do such rules and
representations exist, but their essential structure is already
present at birth and only the fine details are “tuned” by interac-
tion with the outside world. Interestingly enough, developments
of the past decade or two in the world of mathematical system
theory shed some light on these different views.

In the context of minds and machines, we might paraphrase
the central problem of mathematical system theory as:

Given a set of external rules, can we always find a set of internal men-
tal representations and rules such that the internal rules generate the
same behavior as the given external onesf

Under very weak assumptions about the precise forms and prop-
erties of the external rules, the somewhat surprising answer to

256

PARADIGMS LOST

this question is a definite yes! In fact, the result is considerably
stronger, asserting that not only does a suitable set of internal
representations and rules exist, but that this set is unique, once
we impose the additional condition that it be minimal, i.e., that
there be no more representations created than are absolutely
necessary for mediating the behavior specified by the stimulus-
response pattern. In the jargon of system theory, these abstract
mental representations are called states, with the internal rules
usually termed the system’s internal dynamics.

What all this mathematical mysticism adds up to in our lin-
guistic setting can be summarized by the following steps:

A. Given a stimulus-response pattern (external description) of
a system’s behavior, we can always associate with it a mini-
mal set of abstract mental representations and rules that will
act to reproduce the given external behavior.

B. These “states” and “internal dynamics” can be constructed
directly from the stimulus-response pattern.

C. The role of the mental representations is to mediate between
the environmental inputs and the observed behaviors and ac-
tions of the system.

While the foregoing facts seem to deal a strong blow to the be-
haviorist position with its rejection of the very notion of mental
states, there is a practical loophole that needs closing before
Skinner & Co. can be permanently cashiered.

The proverbial perceptive reader will have noted that state-
ments A through C have been couched in terms of “abstract”
mental representations and rules. What this means for real
brains and real minds is not yet clear. The mathematical facts of
life ensure that if we represent real stimulus-response patterns
within the framework of suitable mathematical structures, then
within those structures we can create by mathematical opera-
tions new abstract entities that play the role of internal mental
states. These states, in turn, generate abstract behaviors and ac-
tions in a suitable mathematical space of outputs. The gap that
needs closing is the production of a dictionary that relates these
mathematical structures and abstract mental states to the ac-
tions of real people and their equally real physical brains. In
other words, the question now shifts to the relationship between
the abstract mental states and actual physical states coded into
our neural circuitry. In the jargon of philosophy, we have to
close the gap between mentalism and physicalism.

SPEAKING FOR MYSELF

257

This problem is identical in form to that encountered when we
deal with the mathematical points of Euclid’s three-dimensional
space E3, where each such point is represented by three numbers
measuring its distance in the x, y, and z directions from some
fixed origin. What is the relationship between these purely ab-
stract mathematical objects we call points, and the points of our
real-world space R3, measuring height, width, and depth in the
physical universe? In developing his analytic geometry, Des-
cartes made the astonishing claim that these two sets of points
are identical, i.e., E3 = R3. This assertion stood the test of time
and experiment until Einstein showed it to be only a good ap-
proximation. We’re now in a similar situation with the problem
of relating abstract mental states to real brain states but, alas,
with no Descartes to show the way. As yet, no one has even come
close to providing a plausible argument that closes this gap. But
since I don’t want to start giving away the next chapter’s theme,
let me say no more about the matter here. Let’s return to a sum-
ming-up and the rendering of a verdict on the case of the lin-
guists.

SUMMARY ARGUMENTS

To be absolutely clear on the point to be settled, let’s review the
bidding. Chomsky’s argument is that all normal human children
receive as part of their genetic birthright a unique language-
acquisition device, or language organ. This organ contains a
hard-wired universal grammar, which children use to learn their
native language quickly and effortlessly. The two key points of
contention are whether the language acquisition device is (1) in-
nate, i.e., inherited, not learned, and (2) unique, i.e., specifically
designed for language and not just part of a general problem-
solving apparatus. Tables 4.3 and 4.4 summarize the competi-
tion. As an aside, the reader will note that I’ve included Fodor
along with Chomsky in Table 4.3. The reason is twofold: First of
all, even though Fodor is primarily a philosopher of mind and
not a linguist, his views on the modularity of the mind are com-
pletely in harmony with Chomsky’s; and second, I want to dispel
the view that Chomsky is the only one who holds to the Prosecu-
tion’s case. In fact, a large number of linguists support
Chomsky’s case, but they do so in ways that are so similar to
Chomsky’s that there is no reason to distinguish among them in

258

PARADIGMS LOST

THE LANGUAGE DEVICE
IS INNATE AND UNIQUE!

PROMOTER

ARGUMENT

Chomsky

universal, generative, transformational grammars

Fodor

modularity of mind

TABLE

4.3. Summary arguments for the Prosecution

LANGUAGE IS MAINLY LEARNING

AND/OR NOT INNATE!

PROMOTER

ARGUMENT

Skinner

operant conditioning

Piaget

stages of cognitive development;

interactionism

Sapir and Whorf “language = world”; relativism

Montague

Montague grammar

Sampson

Popperian learning of hierarchical
structures

TABLE 4.4. Summary arguments for the Defense

a broad treatment of this sort. The interested reader may want
to look into the work of some of these Chomskian comrades-in-
arms cited in “To Dig Deeper.”

BRINGING IN THE VERDICT

On the matter of language acquisition, there’s no doubt for me
as to where to place my money: firmly with the Prosecution and
its claims for innateness and uniqueness. In this sense, I’m a
devoted Chomskian. Let me explain why.

First of all, uniqueness. I find it hard to countenance any of
the claims by Piaget, Sampson, et al. that the human language
facility is just part of the general problem-solving and learning
machinery of the brain. It seems to me there’s just too much
empirical evidence against this claim to take it seriously. For
example, why should language acquisition skills mysteriously

SPEAKING FOR MYSELF

259

disappear for most of us in late childhood if the acquisition
mechanism is part of our general learning abilities instead of
being a specialized skill? If I can learn how to dance the tango
or program a computer at the age of forty, why can’t I learn to
speak Russian or French with equal ease if language acquisition
is just a learning procedure like any other? Returning to the
idea of switch settings in the universal grammar, it seems that a
few lucky souls have the ability to change these settings, even in
adulthood. Most of us, however, appear to have these switches
“soft-welded” into place in childhood, and remain prisoners of
our native language thereafter.

Further evidence along these lines is provided by observations
of people suffering strokes or other types of brain injuries re-
sulting in aphasias. If the language facility were as decentral-
ized as the general learning theories suggest, it seems to me that
the unaffected parts of the brain would pick up the slack and
speech impairments would be a lot less prevalent than they actu-
ally are. In this connection, I must say that the ideas of Samp-
son as they relate to learning the hierarchical structure of
language following a Popperian strategy seem appealing. But I
can’t quite accept his claims that the mechanisms invdlved are
just part of a general learning program. So, all in all,
Chomsky’s arguments for uniqueness of the language organ
strike a more responsive chord with me than the claims of his
opponents.

On the matter of innate ness, the scales also seem to swing in
Chomsky’s favor. Without the benefit of some kind of prepro-
gramming, it seems inconceivable to me that children could ac-
quire the basics of virtually any language within their first few
years of exposure, not to mention the capacity to generate sen-
tences never before heard or spoken. I have already mentioned
the case of paraplegics and language acquisition as an example
of the kind of problem that seems difficult for noninnateness
theorists to deal with. The basic problem is to explain where this
language capacity comes from if it’s not basically inborn, and
none of Chomsky’s opponents have presented a case that even
begins to come close to a viable alternative to innateness.

My support of Chomsky’s views on language acquisition
should not be interpreted as a wholesale endorsement of his en-
tire position on languages, especially the ideas supporting the
universal grammar. On this point I have great sympathy for the

260

PARADIGMS LOST

allegations that the universal grammar unfairly and needlessly
underrates the role of meaning. Personally, I lean to a kind of
innate grammar that combines the generative ideas of Chomsky
with the syntax-semantic combination displayed by the Mon-
tague grammars. On balance, it appears to me that Chomsky is
right on target with his notions of modularity and innate ness,
but off course when it comes to the primacy of syntax over se-
mantics. Perhaps the right course is to put his ideas of mind
together with Montague’s ideas of grammar and then sprinkle
on Sampson’s vision of hierarchical evolution. The convergence
of these three streams of thought might, in my outsider’s view,
lead to a theory of language that would stand the tests of both
time and completeness.

Our focal point in this chapter has been the question of lan-
guage and its development within the specific biological machine
we call a human being. The Chomskian verdict says that the
peculiarities of our biological machinery influence not only the
kinds of languages we can speak but, more generally, the kinds
of thoughts we can think. Question: If we had a different kind of
physical structure, in what way might this change the way we
think? In particular, if we were composed of fragments of sili-
con, metal, and plastic connected up like a digital computer,
would we think in the same way we do as humans? For science’s
best answer to this puzzler, read on.

THE COGNITIVE
ENGINE

CLAIM:

DIGITAL COMPUTERS CAN, IN
PRINCIPLE, LITERALLY THINK

THE TURING TEST AND THE CHINESE ROOM

Can a computer think? I mean really think, just like you and
me, with mental states of the same sort we have when we’re slav-
ing over our taxes, daydreaming about next summer’s vacation,
translating the Spanish ads in the subway, or fuming over our
boss’s obvious faults. Is it even faintly plausible that a machine
of metal, plastic, and silicon can literally experience the same
kinds of mental states that we do in these circumstances? If you
think the question’s easy, consider the following two experi-
ments.

262

PARADIGMS LOST

THE IMITATION GAME

Suppose you wander over to your neighborhood university com-
puter center and enter a room whose only furnishings consist of
a chair and a table upon which sits one of the major factotums
of modern life, a video display terminal and its keyboard. At
that moment, a disheveled, malnourished-looking fellow with the
bug-eyed, slightly demented stare of the dedicated computer
hacker appears, informing you that the terminal on the table is
connected either to a similar terminal in another room at which
sits a more or less normal human being of indeterminate sex, or
to a computer that has been programmed to respond to any sort
of question you may wish to pose, provided that it’s expressed in
everyday English. Neither the computer nor the human is under
any obligation to answer your queries truthfully and, to keep
the experiment with reasonable bounds, your interrogation is
limited to, say, twenty questions or maybe an hour’s worth of
questioning. At the end of the experiment, the hacker will return
and you are to tell him whether you think the terminal is con-
nected to a real, live human being or to the computer. The gen-
eral setup is displayed in Figure 5.1.

To get some feel for the kind of probing possible in such a
situation, let’s sit down at the terminal and play this game
awhile. The following silly dialogue might result, with your
questions shown in normal type while the responses from the
human/computer are given in caps.

Well, what should we talk about?

I CAN TALK ABOUT ANYTHING. WHAT WOULD YOU LIKE TO ASK ME?

Let’s start with something easy, perhaps a little mathematics. If I
have a right triangle with sides of lengths 3 and 4, what’s the
length of the third leg?

AH, THE OLD PYTHAGOREAN THEOREM. THE HYPOTENUSE THEN HAS
LENGTH 5, OF COURSE.

Not too bad. At least you know your high-school geometry. What
about some elementary arithmetic? What’s the square root of 147?

JUST A LITTLE BIT OVER 12, BY MY RECKONING.

Pretty vague for a computer, but acceptable for a human. Let’s
try some current events. What are your views on the prospects
for peace in the Middle East?

THE COGNITIVE ENGINE

263

I NEVER TALK ABOUT POLITICS OR RELIGION. BUT IF I DID, l’D SAY THE
OUTLOOK IS BLEAK, ESPECIALLY WITH THE EVER-PRESENT MEDDLING OF
THE AMERICANS AND NOW THE RUSSIANS.

OK, we’ll stay away from politics. How do you fancy the Yan-
kees’ chances for the pennant this year?

WHAT’S A YANKEE?

I see you’re not a sports fan, either. I thought you told me you
could talk about anything.

I CAN CONVERSE JUST LIKE ANY NORMAL PERSON, BUT l’M NOT AN EN-
CYCLOPEEDIA, YOU KNOW.

Not much of a speller either, I see, but no need to get uptight. I’m
just trying to find some common ground for us to have a conver-
sation. How do you feel about the economy? Do you think this is a
good time to invest in stocks?

THE STOCK MARKET IS THE WORLD’S BIGGEST CASINO, BUT MY HUNCH
IS THAT NOW IS A GOOD TIME TO GET OUT OF THE MARKET, NOT IN.

Why do you feel that way?

TOO MANY SPECULATORS AND TOO MANY HAIR-TRIGGER COMPUTER TRAD-
ING SCHEMES THAT CAN WHIPSAW YOU OUT OF YOUR SAVINGS IN A
MATTER OF MOMENTS. SO UNLESS YOU’RE A REAL GAMBLER, l’D STAY ON
THE SIDELINES.

Well, you certainly seem to have definite ideas about finance.
WTiat do you know about the arts? How about composing a little
poem for me?

you’ll HAVE TO COUNT ME OUT ON THIS ONE. I NEVER COULD WRITE.

264

PARADIGMS LOST

Well, I’ve got to leave now, so let me ask you one last question.

Are you a computer or a human?

i’m an intelligent, thinking entity, bye for now, and thanks

FOR COMING TO SPEAK WITH ME.

Imagine you come back to the computer center and play this
game many times. By random guessing alone, you’d expect to be
able to determine correctly whether you are in contact with the
human or the machine half the time, on the average. Imagine
that after a sufficiently large number of plays of the game, your
success rate in distinguishing the machine from the human is
not substantially better than the 50 percent rate from random
guessing alone. Now we ask: Can the machine think? Well, why
not? After all, the only way we have to decide whether or not
other humans are thinking is by interacting with them in much
the same way we interacted with whoever or whatever was at the
other end of the terminal. So if a sequence of such interactions
leaves us unable to separate the computer from the human, then
it seems perfectly defensible to argue either that the machine is
thinking or that humans do not. Since ex hypothesi humans do
think, we must accept that any machine that can fool us in the
above Imitation Game is indeed thinking.

The Imitation Game was originally proposed almost forty
years ago by the British computer pioneer Alan Turing in a
landmark paper on the possibility of constructing intelligent ma-
chines. By all accounts, Turing, who played a central role in
breaking the German Enigma code during the Second World
War, was a somewhat emotionally underdeveloped, otherworldly
character given to offbeat pursuits such as “ run-around- the-
house chess” (in which after you make your move, you get up
and run around the house, and if you get back before your oppo-
nent has moved, you’re allowed an extra move), and the “desert
island game” (a kind of survivalist exercise in which you see
how many chemicals can be produced from household substances
using only homemade apparatus), and the simple passions of
long-distance running, bicycling, and violin scratching. It ap-
pears that Turing’s interest in the idea of a thinking machine
was an outgrowth of his war efforts in cryptography, and
shortly after the war’s end he set down his position, together
with a rather detailed rebuttal to the many sorts of objections
that he anticipated might be offered against the notion. It’s
strong testimony to the basic soundness of his vision that even

THE COGNITIVE ENGINE

265

today, almost forty years later, the fundamental ideas he put
forth are as topical and fresh as the most recent work in the
area, as we shall soon see.

The Imitation Game, or as it’s more commonly termed the
Turing Test, has the virtue of being implementable, in principle,
but unabashedly behavioristic in nature, asserting that the exis-
tence of “thinking” is solely a matter of producing convincing
responses to more or less arbitrary stimuli. By the Turing Test,
any “black box” that does a convincing enough job of imitating
a human being in ordinary conversation would be deemed to pos-
sess genuine intelligence and could (and should) be thought of as
a “thinking entity,” just like our friend in the dialogue. Before
giving our uncritical acceptance to this kind of claim, let’s turn
to the second experiment.

THE CHINESE ROOM

Suppose you find yourself inside an enclosed room whose only
entrance is a door containing a small mailboxlike slot. Inside the
room you find a large number of flashcards upon which are
printed Chinese characters, one per card. You also find a big,
dictionaryish kind of book giving instructions in English as to
how to process the flashcards through the slot. For example, a
typical instruction might read: “If the character ‘squiggle’
comes through the slot, then find the card with ‘squaggle’ and
pass it back outside the room.” Friends outside the room pass in
a sequence of such cards, while you look up the appropriate in-
structions in the book and pass back whatever card is called for.
Now unknown to you (since you understand not one word of
Chinese), the cards that are being passed in form a set of ques-
tions about, say, a current popular film. And the cards you are
called upon to pass back out constitute perfectly sensible, coher-
ent replies to questions about the plot, the actors, the staging,
costumes, and so forth. As far as those outside the room are
concerned, the black box consisting of the room and its contents
displays a perfect understanding of Chinese; however, from
your perspective inside the room, there’s no understanding at
all. You’re just shoving tokens (flashcards) around according to
a set of rules. In short, there is syntax but no semantics.

Now we again ask: Can computers think? Since thinking pre-
sumably involves understanding the meaning of symbols, and

266

PARADIGMS LOST

computers only manipulate symbols according to a set of rules,
the Chinese Room setup leads clearly to the contention that com-
puters cannot think. There is no understanding of the questions
posed in Chinese, just blind symbol manipulation. And without
understanding there are no genuine cognitive states; ergo, no
thinking.

The Chinese Room experiment was proposed by Berkeley phi-
losopher John Searle by way of a counterattack on the adequacy
of the Turing Test as an operational procedure for identifying
objects having genuine mental states. Searle’s claim, of course,
is that your actions inside the room duplicate exactly the func-
tional activities of a computer, and it’s obvious that there’s no
real understanding on your part of the questions being passed
into the room. Whatever understanding exists is present solely
in what’s been “programmed” into the rulebook from the outside,
and the processor (you) has no notion whatsoever of what the
symbols actually mean.

Notice the crucial shift of perspective on the question of
whether or not the black box, consisting of the room and what-
ever may be inside it, possesses actual cognitive states. Looked
at from the outside as called for by the Turing Test, the room
does indeed display every sign of being a thinking being, and we
would justifiably deem it so from our outside, third-person per-
spective. Yet when we take the insider’s first-person stance ad-
vocated by Searle, it’s difficult to see how anyone could take
seriously the idea that the box has internal mental states.

When Searle first published the Chinese Room argument in
1981, the room and its implications met with an outburst of in-
dignation and a variety of denunciations from several quarters
of the artificial-intelligence (AI) community. The well-known AI
advocate and writer Douglas Hofstadter termed the paper “one
of the wrongest, most infuriating articles I have ever read in my
life,” and regarded it as “a religious diatribe against AI.” Simi-
larly, the philosopher Daniel Dennett claimed Searle’s argu-
ments were “sophistry.” We’ll take a look at several of these
arguments later, but for now it’s sufficient to note that third-
person and first-person perspectives lead to flat-out contradic-
tory conclusions regarding the “mentality” of whatever is
shuffling the cards out through the door slot of the Chinese
Room. They can’t both be right, although they could both be
wrong, depending upon exactly how we understand the term

THE COGNITIVE ENGINE

267

“mental state.” If we add to this the fact that humans are in
some sense machines that clearly think, then we’re quickly led to
see that resolution of the possibility of machines’ having legiti-
mate mental states, solely by virtue of their following rules for
formal symbol manipulation, involves sharpening considerably
our ideas of what we mean by a “machine,” a “rule,” a “cogni-
tive state,” and, most important, what we mean by “thinking.”
But before trying to clarify these matters, it’s worthwhile to
pause for a moment and consider why it’s of more than passing
philosophical interest to spend time grappling with such a ques-
tion in the first place.

Contrary to popular belief, researchers claiming the existence
of genuine cognitive states of the human sort in machines do so
neither to undermine cherished psychological, religious, and/or
sociological prejudices surrounding the special position of man-
kind in the universe, nor to demonstrate that man is nothing
more than a machine. The reason for the deep concern with the
seemingly academic question of whether machines have mental
states is distinctly more pragmatic.

Over the past decade or so, the digital computer has provided
the “society of mind” community with an unprecedented tool
for experimentally testing whatever theory of mind one might
fancy at the moment. If you think a neuronal net wired up in a
certain fashion will produce responses only when stimuli occur
in pairs, well, you can just program it into the computer, and
check it out. Or if a colleague claims that language acquisition
involves a particular kind of symbol representation in the brain,
a program can be written to test the proposed theory. So perva-
sive has the digital computer become as a laboratory tool that a
whole new field, cognitive science, has emerged as an amalgam of
psychology, philosophy, anthropology, neurophysiology, com-
puter science, and linguistics, organized around the use of the
computer as a probe for teasing out the secrets of both the brain
and the mind. Consequently, if it can be definitively demon-
strated that no digital computer, no matter how cleverly pro-
grammed, can ever possess mental states of the sort found in a
biologically based human brain, then the computer studies of
mind can be at best simulations of human cognitive processes. On
the other hand, should it turn out that computers can indeed
think just like you and me, then the hand of the cognitive scien-

268

PARADIGMS LOST

tist will be enormously strengthened when he claims that his pet
theory of the mind should be taken seriously, solely because the
computer program’s behavior agrees with the behavior of hu-
mans under similar circumstances. In short, in this case we
could say that the program serves as a model for human
thought, not just a simulation.

As decisions and actions are taken about human beings on the
basis of pronouncements from the psychological community, and
modern life abounds with such actions in every area from decid-
ing university admissions to the determination of who’s crimi-
nally insane and who isn’t, the question of whether or not
machines can have mental states is of practical as well as philo-
sophical importance. Now let’s get back to the question itself.

FORMAL SYSTEMS, MACHINES, AND TRUTHS

Generally when we speak of machines, we have in mind things
like electric motors, drill presses, water pumps, and the like.
These are all devices whose purpose is to act on matter in order
to transform or transport it in some fashion. A computer is a
quite different sort of “machine.” Its purpose is to manipulate
not matter or energy, but rather information. Boiled down to its
essentials, a computer is a machine for transforming one set of
meaningless symbols into another; in short, a device for physi-
cally executing the operations called for by the rules of a formal
logical system. So before we can speak meaningfully about
whether such machines can think, we’ll need a clearer picture of
what constitutes a formal system, and the degree to which the
mental life of humans can be captured by such a system.

FORMAL SYSTEMS

Quite generally, a formal system is nothing more than a set of
abstract symbols, together with some rules specifying how we
can combine strings of such symbols to form new strings. More
specifically, the components of a formal system consist of

• an alphabet composed of a set of symbols, or tokens, such as
the characters [a, b, c . . .) of the Roman alphabet or an even
more culture-free set like {o , 0, A • • •)• Any finite set of these
symbols is called a string. However, most such strings are non-
sense, so we have

THE COGNITIVE ENGINE

269

• a grammar, which is a criterion for determining which strings
are acceptable. Grammatical strings are termed admissible strings
of the system. Finally, to compose a formal system we need

• a set of admissible strings given a priori, termed the axioms of
the system, together with

• a set of rules of inference specifying the allowable ways of com-
bining admissible strings to form new admissible strings.

To fix these very abstract but absolutely essential notions,
let’s look at three everyday examples of formal systems in
action.

Example 1: The Game of Chess. As our first illustration of a for-
mal system, think of the game of chess, where the symbols are
the black and white pieces. The strings of the system are simply
the set of all possible ways the pieces can be arranged on the
board. The grammar is just the specification of all legal posi-
tions that the pieces can occupy on the board (e.g., White King’s
Bishop only on White squares), while there is only a single
axiom, namely, the initial position of all the pieces at the begin-
ning of the game. The rules of inference consist of all legal
moves that can be made at any stage of the game, enabling the
initial axiom to be transformed into a sequence of legal posi-
tions.

The chess example makes it evident that whatever particular
physical properties the pieces and board may possess are irrele-
vant to their role in the game. Thus, it matters not one whit
whether we use ivory or wooden chess pieces, or if the board is
made of stone or plastic, or if the pieces have been formed to
represent agents of the CIA and KGB, or even if we use mate-
rial symbols at all! The only thing that’s important is the ar-
rangement of the pieces in relation to each other and to the
squares on the board, and any abstract symbol strings possess-
ing the right relationships will serve equally well for represent-
ing everything that’s important about the game of chess. It is in
this sense that we say that only the “form” of the symbol
strings is important, not their content, and this is why we term
such systems formal systems.

Example 2: Scrabble. Another board game of universal appeal
that fits into the framework of a formal system is Scrabble. For
those unfamiliar with the game, it is played with a collection of

270

PARADIGMS LOST

small, square wooden tiles, each bearing a letter of the alphabet.
The tiles are placed on a board ruled off into squares much like a
checkerboard, except with far more squares. There is a point
value attached to the letter on each tile, and the objective of the
game is for players to form words by placing the tiles on the
board in much the same fashion as in a crossword puzzle, i.e., by
building upon the words already present. As each word is
placed, points are awarded to the player according to the values
on the tiles used in formation of the word.

The symbols for the formal system describing Scrabble are
just the letters of the alphabet etched onto the individual play-
ing tiles. As in chess, the corresponding formal system for
Scrabble has only a single axiom, which is the initial word placed
on the board by the player who starts the game. But unlike
chess, where the sole axiom is determined by the initial position
of the pieces, which is always the same, Scrabble ’s single axiom
changes from game to game depending upon the choice made by
the first player. The strings of the Scrabble system are just fi-
nite sequences of tiles, i.e., combinations of letters, while the
grammar specifying which strings of Scrabble tiles are admissi-
ble is given by the rules of the game. In general, any string is
admissible if it constitutes a genuine word from the dictionary,
and if the string touches a tile in any other string that’s already
on the board. It is this last condition that ensures that the vari-
ous strings interlock on the Scrabble board in the crisscross pat-
tern of a crossword puzzle. Finally, the rules of logical inference
telling us how to form new admissible strings from old ones are
just the usual rules of Scrabble telling us in what manner tiles
can be added to the board. For instance, one such rule is that the
tiles can be added only vertically or horizontally, not diagonally.

It’s of significance to note here that if you play Scrabble (like
my friend Joe) by introducing your own private dictionary into
the game, different from that employed by the other players,
then you’ll see a different formal system, hence a different game.
This new game may or may not be similar to the original Scrab-
ble, depending upon how similar the new dictionary is to the old,
thereby opening up the possibility for many of the Scrabble
squabbles familiar to the game’s devotees (like Joe’s wife,
Peggy). The point is that any change in any component of the
formal system results in a new formal system. And this new sys-
tem may or may not bear a close relationship to the original.

THE COGNITIVE ENGINE

271

The examples of both chess and Scrabble, as well as other
board games that can be expressed as formal systems like go and
Mah-Jongg, account for part of the fascination that such games
hold for AI researchers. The fact that these games can be repre-
sented by formal systems means, as we shall see, that such games
can be “mechanized” in a precise sense of that term. But before
moving on to consider these matters, let’s first look at another
example of a formal system that is not a board game but per-
haps is even more familiar.

Example 3: Addition. Suppose the symbols of our system consist
of the two characters » and 0. The strings are then just finite
sequences of these two symbols taken in any order. Typical
strings are sequences like 000***** and 0000*******.
All such strings are assumed to be grammatical. Our system will
have the two axioms » and 0, meaning that the single-element
strings » and 0 are assumed to be admissible a priori. We will
allow two rules of inference by which we can generate new
strings from old:

1) S + 0 = 0S and 2) S + * = S*

Rule 1 means that given any string S, we can combine it with
the string 0 and thus obtain a new string consisting of the
string S prefixed by 0. Similarly, Rule 2 says that if we com-
bine S with the string «, then the result is the new string con-
sisting of the string formed by appending the symbol * to S.

Let’s use these rules on the axiom * and see what we get:

S = * (Axiom)

« _ 0# (Rule 1)

0* -> 00* (Rule 1)

00* - 00** (Rule 2)

00*« _» 000** (Rule 1)

In this sequence, each of the strings following the axiom * con-
stitutes what is termed a theorem of the formal system, and the
sequence of application of the rules forms what we call the proof
of the theorem. Thus, the symbol string 00* is a theorem hav-
ing the proof sequence Axiom -* Rule 1 -* Rule 1. Other theo-
rems would have resulted if we had begun with the axiom 0,

272

PARADIGMS LOST

and/or if we had used a different sequence in applying Rules 1
and 2.

So far, the above formal system just gives us a way of gener-
ating grammatically correct strings involving the abstract sym-
bols * and 0. Now suppose we try to attach an interpretation to
these symbol strings in the following way: To each string S, as-
sociate the nonnegative integer [«], where n is the number of
appearances of the symbol * in the string. Thus, the string
000*** and the string *** would both be associated with the
number [3], while the strings 0** and 00** would both be
identified with the number [2], With this interpretation of a
string S, we are able to assign a single integer to each grammati-
cal string of the system. Now if we think of the abstract symbol
0 as standing for our usual notion of zero, it’s easy to interpret
the general Rules 1 and 2 as the ordinary rules of addition, i.e.,

1) [n] + [0] = [«], 2) [«] + [1] = [» + 1]
for every natural number [«].

Thus, the abstract formal system defined solely in terms of the
symbols 0 and * can be modelled by the process of addition of
nonnegative integers — once we make the appropriate interpre-
tation of the symbols and symbol strings. What’s important to
note here is that the symbols 0 and * don’t mean anything until
we pass to the interpretation step; at the level of the formal sys-
tem, they are just symbols or tokens, and the rules of inference
are just prescriptions for shuffling around symbol strings in
order to create new symbol strings. This point is of crucial sig-
nificance when it comes to assessing many of the arguments of-
fered against the idea of a computer actually thinking. At the
formal system level, there is syntax alone; semantics enters only
when the symbols are interpreted, and for a computer to think it
must be possible for the machine to make this transitional step
from the syntax to the semantics. In the Chinese Room experi-
ment, Searle claims this is impossible. We’ll see why later.

The fact that it’s only the form and syntactic structure of the
strings that are important in a formal system accounts for one
of their greatest attractions: They can be about anything. All we
need do is attach some meaning (i.e., semantic content) to the
symbols and presto! Before our very eyes, the system strings
become meaningful statements about the integers or the solar

THE COGNITIVE ENGINE

273

system or the stock market or whatever other interpretation
we’ve given to the symbols. On the other hand, the “meaningless-
ness” of a formal system is also its Achilles’ heel, since the
truths it can express about the real world are entirely deter-
mined by the interpretation injected into the system from the
outside. Thus, the only semantic content that a formal system
can express is there not from inside the system itself, but from
the meaning put into the system from the outside by its user.
This observation accounts for Searle’s claim in the Chinese
Room experiment that “you can’t get semantics from syntax.”
On the surface this argument looks airtight but, as with all mat-
ters of this sort, things are seldom what they seem, and the hid-
den assumptions built into it play an important role in our later
consideration of the objections to the notion of mental states for
machines. For the moment, let’s take a harder look at the ques-
tion of what kinds of truths can be generated by any kind of
formal system.

PROOFS AND TRUTHS

The “truth” or “knowledge” to be obtained from a given formal
system consists entirely of the statements that can be generated
or proved from the system’s axioms by applying the given rules
of inference. Speaking more precisely, a proof sequence in a for-
mal system F is a list of admissible strings Su S2, . . , , Sn such
that each string is either an axiom of F or is obtained from some
of the previous strings by applying the rules of inference. So,
for example, if the system F is the one representing the game of
chess, and the string S , is the sole axiom, consisting of a listing
of the positions of the playing pieces at the beginning of the
game with White to move, then S 2 might be the position of the
pieces after a King’s Pawn opening. That is, S2 is the same as S',
with the single exception that the White King’s Pawn has been
moved forward two squares. A string T is said to be provable in
F if there is such a proof sequence that ends in T, i.e., a se-
quence St — S2 -> • . . . -* T. The set of all provable strings
constitutes the theorems of the formal system F. This setup
should be familiar to each of us as the situation encountered in
our late and unlamented high-school geometry course, where we
started with a handful of elementary, “self-evident” truths
about points, lines, circles, and planes, and proceeded to struggle

274

PARADIGMS LOST

with the rules of logical inference in our feeble attempts to
rediscover a few of Euclid’s ancient truths. Since the purpose of
a formal system is to generate proofs of theorems, we might
think of a formal system as an abstract machine that prints out
the list of theorems provable in the system F.

When it comes to the matter of how powerful a given formal
system F is in its ability to generate a long list of truths, there
are two aspects of the system that bear heavily on the question:
completeness and consistency. Basically, the idea is that we would
like every true statement that can be interpreted using the symbols
of F to be a theorem, i.e., provable, while at the same time being
unable to prove any self-contradictory statements. More infor-
mally, we want F to be able to prove all “ true ” statements, and not
be able to prove any “ false ” ones. So, if 7\, T2 , . . . , is the list of all
theorems provable within F, and P is an interpreted string corre-
sponding to a true statement, then F is called

• complete if P appears on the list Tt, T2, . . . , and

• consistent if P and not-P do not both appear on the list.

Note that the properties of completeness and consistency are
what are termed metamathematical statements about the system
F; i.e., they are not statements (strings) expressible within F,
but rather are statements made about F from the outside, so to
speak.

In terms of the formal system characterizing the game of
chess, the system would be complete if any legal position of the
pieces could be achieved through a legal set of moves starting
from the initial placement of the pieces. The system would be
consistent if a legal position and its negation could not both be
attained. So, for example, if we have the usual legal position
that the White King’s Bishop plays only on White squares, then
any sequence of legal moves that would involve putting this
piece on a Black square would imply the system’s inconsistency.

From the standpoint of machine cognition, it’s of great inter-
est.to understand the difference, if any, between what is “true”
and what is “provable,” since if we could establish the equality

True statements = Provable statements

then we would have gone most of the way toward showing that
all thought processes are just physical manifestations of partic-
ular formal systems. Regrettably for mechanists, things just

THE COGNITIVE ENGINE

275

didn’t turn out this way. We’ll see why in a moment. But first,
let’s pause to catch our breath and summarize in the box below
the impressive array of terminology introduced so far about for-
mal systems.

FORMAL SYSTEMS

alphabet a collection of abstract symbols or tokens used to
form the strings of a formal system

string any finite sequence of symbols (sometimes termed a
formula )

grammar a set of conditions or criteria that distinguish an
admissible string from one that is inadmissible

rules of inference a collection of logical operations that can be
performed on strings to transform one admissible string
into another

axiom a string that is taken to be admissible by definition,
i.e., without proof

formal system an abstract entity consisting of an alphabet,
strings, a grammar, rules of inference, and axioms

proof sequence a finite sequence of admissible strings such
that each string follows from its predecessor by applying
one of the rules of inference

theorem the final, or termination, string in some proof se-
quence

complete system a formal system in which every interpreted
true statement can be proved, i.e., every such string is a
theorem of the system

consistent system a formal system in which an interpreted true
statement and its negation are not both provable, i.e., they
are not both theorems

DIGITAL COMPUTERS

In the crudest terms possible, we can think of a digital computer
as being a device with the capability of storing and changing a
whole lot of numbers. A good analogy would be a general post
office with a large number of post boxes, each box having its own
label or address. We suppose that each box can contain a single

276

PARADIGMS LOST

number. This collection of boxes forms the memory unit of the
computer. Imagine now that we have another device that enables
us to go to any two boxes, remove the numbers that reside in
these boxes, and perform an arithmetic operation upon them,
forming a new number. Such a device is termed the arithmetic
unit of the computer. Similarly, suppose we have another device
that can compare any two numbers and tell us which of the two
is the larger. We call this the computer’s logical unit. In addition
to these units, suppose we also have an input unit enabling us to
place particular numbers into certain boxes, and an output unit
that gives us the ability to look into any box and read its con-
tents. Finally, imagine we have a set of instructions telling us
what boxes are to be looked into, and which further details the
sequence of arithmetic and logical operations to be performed.
This set of instructions is the program. Thus, the way the com-
puter works is first to place a particular set of numbers in some
of the boxes. Next it consults the program to see what the first
operation is to be, goes to the boxes called for by this instruc-
tion, and performs the indicated operation, placing the result in
the particular box that’s specified. It then executes the next in-
struction in the program and carries on in this fashion until it
comes to the end of the program. The computer then employs its
output unit and looks into certain boxes to read out their con-
tents, which we then call the results of the program (in actual-
ity, the input and output operations are also specified as part of
the program and and may be carried out as intermediate steps in
the overall computation). This entire setup can be schematically
depicted in the following diagram:

INPUT

MEMORY

OUTPUT

t i

PROGRAM

ARITHMETIC

AND

LOGICAL

UNITS

In real life, the computer becomes a lot more useful if we can
use it to do more than just perform arithmetic operations on
numbers. In fact, most computers in use nowadays are employed
for things that have little to do with numerical computation, but
rather involve activities like preparing, storing, and retrieving

THE COGNITIVE ENGINE

277

text, creating graphics, monitoring industrial processes, and a
host of other nonnumerical activities. So how is it that we can
arrange for the “number processor” described above to act as a
“symbol processor”? The answer is obvious: Simply code what-
ever symbols we want to process as numbers. In the case where
the symbols we’re interested in are the usual alphanumeric char-
acters of the Roman alphabet, there is a universally agreed-upon
way to associate a number with any of the symbols {A, B, C,
. . . , a, b, c, . . . , 1, 2, . . .}. This labeling of symbols with
numbers is termed the ASCII (“As-key”) code, and it works in
the following manner.

The basic unit of storage in a modern computer is a unit
called a byte, which consists of a string of eight binary digits or
bits. Thus, every address location in the computer memory can
store a single number consisting of a string of eight bits. In the
ASCII coding scheme, the first bit in each byte is reserved for
various sorts of internal bookkeeping chores, leaving seven bits
free to code alphanumeric quantities. So there are a total of 27
= 2x2x2x2x2x2x2 = 128 different quantities that
a single byte could encode. Here are a few examples of how the
ASCII code allows us to represent alphabetic and numeric sym-

bols:

SYMBOL

ASCII CODE
1000001
1001101
1001001
0100001
0100000
0111111

u (blank space)
?

Thus, in ASCII the sentence “I AM!” would be translated into
the byte string

I u AM! = 1001001/0100000/1000001/1001101/0100001

while the interrogative “I AM?” would be the sequence

I u’ AM? = 1001001/0100000/1000001/1001101/0111111.

Using this kind of coding, we can then employ the computer
memory locations to store individual alphanumeric symbols as

278

PARADIGMS LOST

well as numbers, and arrange things so that the computer can be
used not just as a “number cruncher” to do arithmetic calcula-
tions, but also as a symbol processor to manipulate nonnumeric
quantities. This kind of coding scheme enables us to see how a
computer might be used to determine mechanically the theorems
of a formal system. In fact, we can make an argument to show
that symbol manipulation in a computer according to a specific
program is exactly the same thing as the determination of the
theorems of a particular formal system. Let’s see why.

In a digital computer, the symbols of the formal system are
just the elements 0 and 1, while the grammatical strings are all
those binary sequences whose length equals the word length in
the computer. This is set by the computer hardware design, typi-
cally two or four bytes for a standard personal computer. The
axioms of the formal system are the strings that encode the in-
puts fed in at the beginning of the calculation, while the rules of
inference are just the statements composing the program that
operates on these input strings (axioms). Thus, every computer
programmed to deal with a particular kind of problem is a for-
mal system in exactly the sense described earlier.

By a result due to the same Alan Turing, the inventor of the
Imitation Game, the converse is also true: Every formal system
is equivalent to a suitably programmed digital computer. In
fact, Turing proved much more. He showed the existence of a
universal computer, which, given enough memory and time, can
simulate any computer, and that any formal system could be
modeled by running an appropriate program on this universal
computer, or Turing machine. Thus, an IBM PC could simulate
the behavior of a Cray YM-P (but verrry slowly, since compu-
tational speed is hardware-dependent). Further, the so-called
Turing-Church Thesis states that every computable quantity
(roughly speaking, every output that can be obtained as the re-
sult of following a program) can be computed on a Turing ma-
chine. So the problem of mental states for machines now becomes
equivalent to the question: Are human cognitive processes (i.e.,
is thinking) representable by a formal system? In other words,
do all human cognitive processes involve just manipulating a
collection of abstract symbols according to a set of rules? If so,
what are the symbols and rules; if not, what’s missing? The an-
swers hinge critically upon an understanding of just what kinds
of knowledge or truths are forma lizable in the sense that they are
the theorems of some formal system.

THE COGNITIVE ENGINE

279

GODEL’S THEOREMS

The most influential mathematician of the early part of this cen-
tury was the German David Hilbert, who thought that all possi-
ble mathematical truths could be captured within some formal
system, and who actively promoted the formalist school of math-
ematics devoted to a rigorous proof of this contention. Formalist
hopes were permanently blown away in 1931 by Kurt Godel, who
astonished the mathematical (and philosophical) world by prov-
ing that for any formal system JF that is (1) finitely describable,
(2) consistent, and (3) strong enough to prove the basic facts
about elementary arithmetic,

I. P? is incomplete,

and

II. PF cannot prove its own consistency.

Godel’s theorems show that every formal system is subject to
inherent limitations on the amount of “truth” that we can ex-
pect to squeeze out of it. Godel I states that no formal system J*"
is capable of deciding every statement that can be made about
the natural numbers. Thus given a formal system Jr, there is a
statement SP about the natural numbers that can be made (and
even seen to be true), but that cannot be proved in PF\ moreover,
if we extend P? to include SP (for example, by including SP as
one of the axioms of a new system J*”), then there is a new true
statement SP' that is not provable within PP’. Also, if PF is to
embody a correct description of all mathematical truths, we
would expect the consistency of PF to be readily apparent and a
fairly easily provable fact. Nevertheless, Godel II tells us that
this just isn’t so: Even if PP is consistent, we can’t use PP to
prove this fact. Actually, this result can be even further
strengthened to the statement that there exists no constructive
procedure that will suffice to prove the consistency of PP .

These very abstract results can be seen more clearly if we in-
terpret them as “just” special cases of an even stronger result of
Gregory Chaitin on the limitations of formal systems (or,
equivalently, Turing machines) in their ability to cope with
complexity. Specifically, suppose we have a string composed
of 0’s and l’s. Some such strings are intuitively “simple,”
like 0000 ... 000 or 1010101010101010. Others, like
001001110101010011010, have no apparent pattern and look

280

PARADIGMS LOST

“complicated.” The great Russian mathematician Andrei Kol-
mogorov and, independently, the American Chaitin had the idea
of characterizing the complexity of such a string by using the
notion of a Turing machine and a program for producing the
string. In particular, they argued that if the program required to
produce the given string was of about the same length as the
string itself, then such a string would be more complex than one
that could be produced using a relatively short program. Thus,
for example, the string consisting of all 0’s can be produced by
the simple program: “Start with 0 and continue in this way for as
many elements as are in the given string.” Thus no matter how
many 0’s are in the string, we can always produce the given
string with this simple, relatively short program. On the other
hand, the “complicated” string above seems to have no program
appreciably shorter than just instructing the machine to write
out the string itself. Using this line of reasoning, Kolmogorov
and Chaitin defined the complexity of a string as being the length
of the shortest program needed by a universal Turing machine to
produce the string. Since as we have seen, a program can also be
described by a finite binary sequence, there is no ambiguity here
as to which of two given programs is shorter than the other.

With the above notions in mind, in 1965 Chaitin proved the
following remarkable result: If & is a formal system that is
(1) finitely described and (2) consistent, then there is a number
x such that the system ^ cannot prove that there are any bi-
nary strings with complexity greater than x. In other words,
any formal system J*" is limited in its ability to determine the
complexity of an arbitrarily given binary string. But since there
are infinitely many strings of arbitrary complexity, it must cer-
tainly follow that there are strings of complexity greater than
any arbitrary, but fixed, number x. But & is unable to prove
this fact, so it must be that & is incomplete. Thus, using Chai-
tin’s Theorem we are able to deduce Godel’s Incompleteness
Theorem as a simple corollary.

Rumor has it that Hilbert was livid with rage when informed
of Godel’s results, perhaps not surprisingly, since having years
of work, as well as one’s philosophical way of life, destroyed vir-
tually overnight is a bitter pill to swallow. As one might suspect,
the proofs of Godel’s and Chaitin’s incompleteness theorems are
much too technical to enter into here, but the underlying trick
that makes the magic work is to find a way to mirror the meta-

THE COGNITIVE ENGINE

281

mathematical properties of completeness and consistency within
the system & itself. The basic idea shows up already in the fa-
mous Liar’s Paradox, illustrated by the statement

THIS SENTENCE IS FALSE.

Here we can interpret the expression at two levels: the level of
the words in an ordinary English sentence, and a higher level
referring to the meaning of the sentence. Thus, the sentence can
speak about itself in a semantic sense by using symbols and
rules at a purely syntactic level. The way Godel achieved this
kind of self-reference for formal systems is indeed tricky and
devious, just the kind of argument one might expect from a man
who, according to mathematical folklore, agonized for weeks
while studying the U.S. Constitution for his citizenship exami-
nation because he thought he had discovered logical contradic-
tions built into it by the Founding Fathers of the republic!

The key ingredient in Godel’s proof of the foregoing results
was the construction of a string G that represented a mathemati-
cal way of saying “I am not provable.” Then if it were possible to
prove G, the string G would be false and the formal system con-
taining G would be inconsistent; on the other hand, if G could not
be proved, then we would see that G is true but impossible to
prove using the rules of inference of the formal system; i.e., the
system is incomplete. Godel’s genius was to prove that such a
Godel sentence G could be found for any formal system that
was sufficiently rich to contain the usual rules of arithmetic.

Figure 5.2 shows a schematic version of Godel’s result in
“logic space,” where the enclosed box represents all possible logi-
cal statements that can be made. Let the box initially be colored
completely gray. Suppose M is a given, finite mathematical the-
ory, i.e., a formal system. Using M, we are able to prove some
logical statements true and falsify others. Let the true state-
ments be colored white, and the false ones black. Thus, starting
with the theory M, we gradually change the color of the logic
square from gray to a mixture of black, white, and gray. What
Godel says is that there is no theory M that will enable us to
remove all the gray. In other words, there will always be some
statement of the type denoted in the figure by CM, which is for-
ever doomed to lie in the twilight zone of logical grayness. Of
course, different theories remove different regions of gray, but
no single theory, or combination of individual theories, can re-

282

PARADIGMS LOST

move it all. One of the crucial questions for the proponents of
thinking machines is to address whether or not the gray area
that remains is accessible to humans, but not to computers.
We 11 return to this point in detail in a later section.

Now that we’ve taken a high-altitude flight over the territory
of formal systems, truths, proofs, and Godelian logic, let’s try to
bring these purely logical ideas into contact with machines and,
m particular, the digital computer in an attempt to see what
these stratospheric mathematical abstractions have to do with
what is computable by such devices.

MACHINE STATES AND COGNITIVE TRUTHS

The preceding discussion has shown that each memory address
in the computer can hold exactly one byte of information. Thus,

THE COGNITIVE ENGINE

283

we can specify the entire state of the computer’s memory at any
given moment by giving a list of what symbols are currently
being stored in each of its memory locations. It can be shown
that the other functional units in the computer can also be char-
acterized by their own byte patterns, so that we can speak of the
computer’s state at any time as a list of the byte pattern that
is currently present in each of its basic units. Hereafter, when-
ever I speak of the state of the machine, or, more compactly,
the machine state, it will mean such a list consisting of the byte
pattern currently residing in the machine’s memory unit, its log-
ical unit, arithmetic unit, and so on. Since execution of the pro-
gram will result in a change of these states as time unfolds, we
can also think of the computer’s state history as being a listing
of its successive states over the entire time history of the compu-
tation.

It is commonly held that cognitive thought in humans is some-
how associated with the various electrochemical activities taking
place in the neurons inside the brain. To oversimplify slightly,
we can think of a neuron’s state at any moment as being either
“on” or “off,” depending upon whether the neuron is firing at
that moment or not. At the neuronal level, a listing of the state
of each neuron constitutes what we can call a brain state at that
moment. Somehow (nobody really knows how) these brain states
give rise to the mental states that we associate with thinking.
Thus, there is some kind of correspondence between physiologi-
cal brain states and a set of abstract states that represent ordi-
nary cognitive notions such as our mother’s face, a car, a pain,
or a sunny day. In what follows, we shall use the general term
cognitive state or mental state for these abstract quantities. If
there is any content whatsoever to the claim that computers can
literally think, or at least think like you and me, then there must
be a way in which the computational states of the machine can
be meaningfully associated with these mental states of human
thinkers. So far, a detailed account of this association remains
but a gleam in the eye of the AI aficionados, and there are many
who claim that no such connection between the machine and the
mental will ever be made. Nonetheless, the resolution of the
thinking machine debate ultimately resides either in producing a
convincing map between the two, or in proving that it does not
exist. In short, the problem is whether or not it’s possible to
remove the question marks in the diagram:

284

PARADIGMS LOST

machine

brain

X, X

cognitive

states

This theme is our leitmotiv for the remainder of the chapter.
Before listening to the competing claims, let’s briefly return to
some of Godel’s thoughts on the matter.

It’s evident that Godel’s results have a profound bearing on
the issue of thinking machines, since they appear to imply that
there exist truths that can be known but that cannot be captured
by any formal system and hence cannot be obtained by any kind
of computation. There is considerable controversy over the
meaning of Godel’s theorems for the question of artificial intelli-
gence, and we’ll examine some of the competing arguments later.
But for now it’s of interest to hear just what Godel himself
thought about this question. Unfortunately Godel was rather re-
clusive and secretive, especially in his later years, and his only
published statement on the topic comes from a lecture delivered
to the American Mathematical Society in 1951:

The human mind is incapable of formulating (or mechanizing) all
its mathematical intuitions, i.e., if it has succeeded in formulating
some of them, this very fact yields new intuitive knowledge, e.g.,
the consistency of this formalism. This fact may be called the “in-
completability” of mathematics. On the other hand, on the basis of
what has been proved so far, it remains possible that there may
exist (and even be empirically discoverable) a theorem-proving
machine which in fact is equivalent to mathematical intuition, but
cannot be proved to be so, nor even be proved to yield only correct
theorems of Unitary number theory.

Thus Godel leaves open the possibility of the existence of a theo-
rem-proving machine, and even concedes that it may be possible
to discover such a machine by empirical investigation. However,
he then throws a wet blanket on the whole business by saying
that if we ever find such a machine, it will be beyond our powers
to prove that it constitutes a Universal Truth Machine.

We began this section by trying to get a more precise feel for
what we mean when we speak of a “machine,” and ended up
taking off into the stratosphere of formal systems, undecidable
propositions, universal computers, and the like. So let’s try to
summarize the situation thus far. Henceforth, when we speak of

THE COGNITIVE ENGINE

285

a machine we will be talking about a universal computer (a Tur-
ing machine), which, by the Turing-Church Thesis, is capable of
computing anything that can be computed. Furthermore, we saw
that every such Turing machine is equivalent to a particular
formal system, which means that the theorems of the system co-
incide with the quantities computable by the Turing machine.
Finally, Godel’s theorems told us that every such system, and
hence every such machine, is subject to inherent limitations on
the quantity of truth that we can extract from it. Therefore, as
indicated above, the problem of whether or not such machines
can “think” now comes down to a more detailed consideration of
how we can associate “cognitive states” with the “computational
states” of such a machine, and of the connection such cognitive
states have with everyday, garden-variety human thinking and
with the electrochemical activities going on in the brain.

“STRONG” YS. “WEAK” AI, BRAINS,

AND MINDS

By informal consensus, the birth of artificial intelligence as a
recognizable intellectual undertaking can be pinpointed to the
summer of 1956 at Dartmouth College, where John McCarthy,
then a member of the Dartmouth Mathematics Department, con-
vinced the Rockefeller Foundation to fund a summer study on
“the conjecture that every aspect of learning or any other fea-
ture of intelligence can in principle be so precisely described that
a machine can be made to simulate it.” Along with McCarthy,
who now heads the AI Laboratory at Stanford University and
who bears responsibility for coining the term “artificial intelli-
gence,” others at that historic Dartmouth workshop included
Marvin Minsky, head of the MIT AI Laboratory; Claude Shan-
non, inventor of information theory; Herbert Simon, Nobel lau-
reate in economics from Carnegie-Mellon University; and
Arthur Samuel, developer of the first championship-caliber
checkers-playing program, as well as a half-dozen others from
academia and industry who shared the vision that perhaps a ma-
chine could be made to perform human functions that previously
were thought to require intelligence.

It’s of interest to note that even at this dawning of the Age of
AI, the manifesto of the Dartmouth study was already madden-

286

PARADIGMS LOST

ingly vague as to whether or not the participants actually shared
the belief that machines might one day actually think or would
only behave as if they were thinking, each possibility being left
open by use of the word “simulate.” Written and oral accounts
of the meeting support both positions, some of the participants
being engaged in studies of networks of artificial neurons that
they hoped would, in some sense, mirror the biological neurons
of the brain, while others at the meeting were much more inter-
ested in the construction of programs that would behave in an
intelligent fashion, regardless of whether or not the principles
the programs employed bore any resemblance to the way a
human brain would do things. This split between the paradigms

Thinking = The way the brain does it

and

Thinking = The results the brain gets

persists to this day, dividing the AI community into what has
been termed the strong and weak schools of AI.

For purposes of even understanding what the question of
whether machines can think means, it turns out to be of value to
refine the “strong” versus “weak” dichotomy just a bit accord-
ing to a scheme proposed by the philosopher Keith Gunderson.
He identifies the following versions of AI:

• Strong AI, human: Whatever kinds of cognitive states ma-
chines might have, those states are functionally (although, of
course, not physically) identical to those found in the human
brain.

• Strong AI, nonhuman: The kinds of cognitive states found in a
machine are not functionally identical to those in the brain
and hence cannot be used to model human thought processes.

• Weak AI, sim-human: A computer can simulate human cogni-
tive processes, but there is no particular correlation between
the computer states and the cognitive states of the brain.

• Weak AI, sim-nonhuman: A computer can simulate the cogni-
tive processes in a nonhuman mind (e.g., a frog, a dog, an
ant), but the states of the machine may or may not be related
to those in the nonhuman brain.

• Weak AI, task, nonsim: The computer can perform tasks that
previously required intelligence, but there is no intelligence re-

THE COGNITIVE ENGINE

287

quired of the machine, whose states have nothing whatsoever

to do with cognition, human or otherwise.

It’s important here to understand the distinction between two
states’ being functionally equivalent, and being physically iden-
tical. The easiest way to see this difference is to imagine we had
a correspondence between, say, three cognitive states <7,, C2, and
C 3, and three machine states Mu M2, and M,. These states are
clearly not physically identical since the machine states are just
patterns of 0’s and l’s imprinted within some silicon chips, while
the cognitive states are connected with the chemical concentra-
tions and electrical pattern in a brain. However, these two se-
quences of states would be functionally equivalent if, for
example, whenever we saw the machine pattern M , -» M, -> M2
it always corresponded to the cognitive pattern C2 -* C, -» C2.
In this case, we would say that the states M , and C3 were func-
tionally identical because they played the same functional role in
the corresponding sequences; i.e., they were always the middle
state of the three-state sequence.

As far as genuine machine thinking goes, the only category
that counts is the first: strong AI, human; everything else, while
undoubtedly technically challenging and economically reward-
ing, is pretty much devoid of any real intellectual or philosophi-
cal appeal, at least as far as the thinking-machine question goes.
This may come as a surprise to many in view of the recent
brouhaha generated by the media (and various self-serving
members of the AI community), extolling the wonders of the so-
called expert systems being developed in AI labs from Massa-
chusetts to Tokyo, describing the robots waiting just around the
corner to satisfy your every desire (or take your job), and pro-
claiming the need to pour more good money after bad to keep
pace in the “thinking machine race” with the Japanese. And this
is not to mention the venture capitalists/entrepreneurs and their
computer-fixated associates, who are running around doing a
good Keystone Kops imitation while trying to capitalize on the
public’s gullibility over the cognitive capacities of machines.
This whole deplorable situation can be traced to a handful of
programs demonstrating some progress in the last and intellec-
tually feeblest category on our list: weak AI, task, nonsim.
Progress in this category sheds about as much light on thinking
as the flight mechanism of birds shed on the development of the

288

PARADIGMS LOST

airplane. So henceforth, when we speak of cognitive states for
machines, we will be referring to the kinds of states understood
to be in our first category: strong AI, human.

Needless to say, no one has ever produced an unassailable ar-
gument showing that the internal states of an appropriately pro-
grammed digital computer are functionally identical to the
states existing between your ears when you’re engaged in ogling
that new Mercedes, poring over the seemingly endless menu at
the neighborhood Chinese restaurant, juggling your expense ac-
count to cover a session at the craps table in Vegas, enjoying a
Bach fugue, or performing any of the other myriad activities
that, in some sense, we call thinking. Nevertheless, this is our
problem. And as a result of our deliberations thus far, we can
finally state the “Can machines think?” question in more or less
final form:

TURING MACHINE VERSION
Can an appropriately programmed computer
display strong AI, human f

FORMAL SYSTEM VERSION
Are all human cognitive states functionally equivalent to
the admissible strings of some formal system ?

It’s clear that Alan Turing, a computer scientist and logician,
would answer the question with a resounding yes, while John
Searle, a philosopher, would give an equally strong negative
reply. This separation between the “scientists” and the “human-
ists” is typical of the way the deep-thought industry seems to
have divided itself on the matter, but the reasons for taking
these positions are manifold and diverse. But before entering
the courtroom of scientific debate and listening to the competing
arguments, let’s first hear the thoughts of John von Neumann,
who spent the final years of his life reflecting on the problem of
mechanical thought.

Von Neumann, a banker’s son from Budapest, was one of the
few true geniuses of the twentieth century. Before his untimely
death in 1957 from bone cancer (most probably induced by radi-
ation exposure suffered while observing the hydrogen bomb tests

THE COGNITIVE ENGINE

289

at Bikini atoll in the early 1950s), von Neumann made funda-
mental contributions to the theory of logic, quantum mechanics,
meteorology, game theory, economics, and functional analysis.
Important as this work is, there is little doubt now that
von Neumann’s most lasting contribution will be his central role
in the development of the digital computer, particularly the idea
of the stored program. As an outgrowth of his work on the the-
ory of computation, von Neumann became interested in the logi-
cal structure of machines, producing the first proof of the
possibility for a self-reproducing machine, as we detailed in
Chapter Two. In this effort, he anticipated the later work of
Watson and Crick on the dual role of information in cellular
DNA, identifying the need for information to be used in both an
interpreted and noninterpreted form if self -reproduction were to
take place in any sort of organism, biological or otherwise.

Oddly enough, despite his clear understanding of the distinc-
tion between the functional activity of biological organs and
their material construction, von Neumann tended to be some-
what skeptical about the possibility of a computer’s duplicating
the activities of the human brain, primarily because he found it
difficult to see how the physical hardware of the computer could
ever be made to mimic the complexity of the brain. In his last
published work, the incomplete text of his Silliman Lectures at
Yale, von Neumann devoted most of the volume to a detailed
comparison between the hardware of the brain (the neurons,
axons, synapses, and so forth) and the hardware of the com-
puter (flip-flop circuits, switching speeds, reliability, etc.),
taking considerable pains to point out the several-orders-of -mag-
nitude difference between the two in information-processing ca-
pability. But there is virtually no mention of the fact that
computers and brains, despite their vastly different physical
compositions, carry out exactly the same kind of information-
processing functions. It’s as if one were examining a grandfa-
ther clock and a digital watch and were puzzled over the fact
that one was made out of wood and brass while the other was
formed from plastic and quartz, ignoring the fact that they both
performed exactly the same timekeeping function. There are es-
sential differences in the design and construction of the two ob-
jects, but functionally they are indistinguishable.

While von Neumann never actually came out and stated that
he thought a computer could not duplicate the brain, his writ-

290

PARADIGMS LOST

ings strongly indicate that he felt that way, and that the com-
puter could never really mimic the brain because it just wasn’t
made from the right stuff. In other words, when it comes to
human-style cognition, hardware counts. In the battle between
the philosophers and the scientists over thinking machines, this
same argument surfaces again as one of the pillars upon which
the “Computers Can’t Think” school of thought bases its case.
But we’re getting ahead of our story, so with the above prelimi-
naries in hand it’s time to drag out the scales of justice and
listen carefully to the Prosecution and the Defense in an at-
tempt to weed out the few facts from the polemics and hype, and
come to some position on the issue of machine cognition. As in
all trials, we start with the Prosecution.

TOP-DOWN SYMBOL CRUNCHING

Herbert Simon, winner of the 1978 Nobel Prize for economics, is
a soft-spoken, slightly graying man, whose trim figure belies the
fact that he is now in his early seventies and still one of the most
active practitioners of the “artificial intelligentsia’s” arcane art.
His Nobel-winning work was for pioneering techniques aimed at
understanding behavior in organizations and the planning of in-
dustrial activity, originating many of the concepts that we now
know under the rubric “management science.” Somewhat less
well known to the general public is his lifelong interest in the
ways of human thought processes, and the possibility of captur-
ing these principles in computational algorithms. Now Simon is
not noted as being a man of bombast or hyperbole, so one can
imagine the shock when in January 1956 he returned from the
holidays to announce to his class at Carnegie-Mellon University
that “over Christmas, Allen Newell [his colleague at CMU] and
I invented a thinking machine.” By this he meant that he and
Newell had developed a computer program that displayed behav-
ior they considered to be “thinking.” Edward Feigenbaum, now
a well-known exponent of the “expert system” school of AI, was
a student in that class, and his reaction to Simon’s bombshell
was what one might expect: “What do you mean by a thinking
machine?” What Simon, Newell, and their co-worker, J. C.
Shaw of the RAND Corporation, meant by a thinking machine
defines what we can now term the top-down approach to achiev-
ing mechanical thought.

THE COGNITIVE ENGINE

291

Put crudely, the top-down thesis is that human thought pro-
cesses take place as a result of rule-based symbol processing in
the brain. Thus, just as we can go into a chemistry lab and put
atoms of various types of chemicals together according to the
Mendeleev Table, so forming more complicated compounds hav-
ing new properties entirely absent from the individual compo-
nents, the brain can put together the “atoms” of thought (the
symbols) according to various rules, thereby generating the mul-
titude of cognitive states we call thinking. Here we see the prob-
lem of correspondence between cognitive states, brain states, and
machine states in its purest form. The top-downers blithely for-
get about brain states altogether, and just assign various ma-
chine states to cognitive states, much in the same manner that
we earlier assigned ASCII codes to alphanumeric symbols. A set
of rules (usually termed a semantic network or conceptual depen-
dency graph ) telling how these machine states can combine with
each other is then postulated, and the resulting machine states
are “decoded” to give an interpretation of the computation in
terms of cognitive concepts. This, in a nutshell, is the strategy of
the entire top-down approach to AI.

As a simple illustration of the foregoing ideas, here’s a con-
ceptual dependency graph for the idea “John bought a car”:

JOHN

A trans <-

MONEY

-t

JOHN

SOMEONE

<->

A trans <-

CAR

-C

SOMEONE

JOHN

In the diagram, ATRANS refers to the transfer of an abstract
entity, in this case ownership of the car and the money. Many
top-down advocates think that most of our everyday acts can be
broken down into a dozen or so primitive actions, like PTRANS
for the transfer of a physical object and MTRANS for the
transfer of information. The claim (or hope) is that these primi-
tive actions form a language for the representation of meaning
in a computer, with the idea being to code each of these actions
and associated mental states by certain computational states of
the machine, and then put in the rules by which these primitives
can interact to form more complex sorts of activities.

The very first working program of this type, the one Simon

292

PARADIGMS LOST

announced to his mathematical modeling class in 1956, was
called Logic Theorist in reference to its ability actually to gener-
ate proofs for many of the theorems in Alfred North Whitehead
and Bertrand Russell’s magnum opus Principia Mathematica.
Amusingly, Simon and Newell participated in the historic 1956
Dartmouth summer meeting, even producing a working version
of Logic Theorist for demonstration. But the import of this semi-
nal achievement seems to have been lost on the other partici-
pants at the meeting, who more or less ignored what amounted
to the first working computer program to have displayed any-
thing approaching real intelligence.

The underlying principle that Logic Theorist and its successor,
General Problem Solver, employed is a form of heuristic reason-
ing called means-end, analysis. Basically what this involves is not-
ing that when we have a problem to solve, we always start with
(1) a given initial state (data, premises, and so on), (2) a desired
terminal set of states (goals), and (3) a set of operators that can
transform one state into another. The task then becomes to find
a sequence of operators that will transform the initial state into
the terminal set. Simon and Newell supplied their programs with
two kinds of heuristics:

• Procedures for detecting significant differences between two
states

• Rules of thumb about which operators typically reduce differ-
ences between various kinds of states

The solution principle is then clear: Detect some difference be-
tween the initial state and the terminal set; apply some operator
that ordinarily reduces such a difference; if the resulting state
doesn’t differ from the terminal set, stop; otherwise try the same
procedure, but now from the new state.

Example: The Three-Coin Problem. To see how this kind of anal-
ysis works, consider the well-known Three-Coin Problem, in
which we have three coins, each of whose initial position can be
either heads (H) or tails (T). The goal is to transform the initial
configuration into one for which all of the coins are showing ei-
ther H or T, i.e., the goal states are HHH and TTT. For any
given state, there are three possible operators: “turn the first
coin over,” “turn the second coin over,” and “turn the third
coin over.” A move corresponds to the choice of one of these
three operators, and a solution to the problem is a sequence of

THE COGNITIVE ENGINE

293

three moves that will transform the initial state into one of the
goal states.

If we designate the three operators as A, B, and C, corre-
sponding to turning over the first, second, or third coin, respec-
tively, then Figure 5.3 shows the sequence of possible moves that
can be made in this game. Notice from the diagram that it is not
possible to move from the state HTT to the goal state TTT in
exactly three moves.

The solving of logical puzzles, the playing of simple games like
tick tack toe, and a variety of other heuristic search activities
typify what we term automatic formal systems. These are formal
systems that work by themselves in the sense that in their nor-
mal mode of operation, they automatically manipulate the for-
mal symbols of the system according to the system’s rules. All of
the Simon and Newell work on top-down computer cognition can
be classified under the heading of such automatic formal sys-
tems. Unfortunately, several years’ worth of experimenting with
automatic formal systems has led to the sad conclusion (one of
many, of course) that, rather than demonstrating that human
thought is really just formal symbol manipulation in disguise,
what the Simon and Newell exercises show is that game playing,
theorem proving, and the like can be done well without anything
even approaching the full spectrum of human intelligence. In
short, programs like Logic Theorist can produce intelligent-look-
ing results in a very restricted domain, but once out of that do-
main there’s a Grand Canyon-sized chasm separating them from
what anyone would even charitably call thinking.

As an amusing indicator of the nature of the gap that remains
to be bridged between rule-based symbol-manipulation programs
and everyday thinking, some years ago an effort was made to
produce a Russian *—> > English translation program that could
take a text in one language and produce at least a rough transla-
tion into the other, the goal being to relieve a human translator
of the drudgery of doping out the gist of the text so that his
time could be more profitably spent polishing the machine ver-
sion for final consumption. The basic idea was to program a
large vocabulary and the grammar from each language into the
machine, give it a few rules and idioms, and then turn it on. The
immensity of the task was quickly brought out when the simple
phrase “out of sight, out of mind” was translated back as “blind
and insane!”

The nature of the difficulty was identified by the well-known

294

PARADIGMS LOST

FIGURE 5.3. The possible moves in the Three-Coin Came

logician and philosopher Yehoshua Bar-Hillel’s claim that a
computer would never be able to distinguish between the phrases
“the pen is in the box” and “the box is in the pen,” where “pen”
in the second case would immediately be understood by any
human to refer to a baby’s playpen. To make such a distinction,
Bar-Hillel asserted, the computer would need not only a dictio-
nary and grammar, but a universal encyclopedia containing a
vast amount of knowledge about the world, the kind of knowl-
edge that we humans take for granted and routinely acquire as
we stumble through life. Somehow this knowledge must be given
to a machine if it’s to act like a thinking agent, at least from a
human’s perspective.

The natural-language-processing problem illustrates in the

THE COGNITIVE ENGINE

295

starkest possible terms the major difficulty with top-down sym-
bol processors: They just ain’t got no common sense! There’s no
way that programs of the Simon and Newell stripe are ever
going to “think” in the way humans do until a way is found to
code knowledge of the world into the formal symbols that the
computer operates with. In retrospect, it’s easy to see that when
we perceive intelligently, we never perceive an object, but rather
a function and a context. So if I show you a key, you never think
of it as just a machined piece of metal; rather you see it as an
object that performs the function of unlocking something, per-
haps a door, a safe, a car, or whatever the context suggests. It’s
this kind of knowledge that a computer needs if it’s going to
think top-down style.

The past decade or two has seen a number of disparate at-
tempts to deal with the common sense acquisition problem for
top-down AI. Let’s take a glimpse at a couple of the more promi-
nent efforts.

MICROWORLDS

A procrustean approach to giving computers common sense
about the world is simply to fence off most of the outside world
and let the computer have access only to a very severely re-
stricted universe whose features, idiosyncrasies, folkways, and
mores can be spelled out in painstaking detail and then given to
the computer in some sort of digestible form. For example, Mo-
nopoly is a microworld in which the aspiring real-estate tycoons
never have to worry about contingencies like fires, wars, dead-
beat tenants, civil action suits, and the zillions of other annoy-
ances that plague owners of pieces of real-world real estate.
Board games like chess, go, and checkers are other micro worlds
of this type.

Probably the best-known microworld program is SHRDLU, a
block world put together by Terry Winograd in the early 1970s.
This universe consists of a few imaginary blocks of various sizes
and shapes, strewn about on a flat surface. Figure 5.4 shows
SHRDLU' s world. The blocks may be colored and cast shadows,
but they never have any other physical properties beyond their
geometric shapes and dimensions. SHRDLU knows all there is
to know about this microscopic universe, and is able to converse
in a seemingly intelligent fashion when queried about the world

296

PARADIGMS LOST

or asked to perform certain acts such as placing one block atop
another, or picking up a block and moving it to a different loca-
tion.

Despite what appear to be intelligent dialogues between
SHRDLU and its inquisitors, the program has a variety of fatal
deficiencies as a cognitive entity: (1) SHRDLU never initiates
any actions but only reacts to queries put to it; (2) the program
has no motivational goals whatsoever, other than the goals intro-
duced by inquiries from the outside; (3) the main problems of
perception and action involve capturing the interface between
symbolic cognition and real objects. But SHRDLU’s “world” is
already symbolic, so it doesn’t address this interface at all. But
these difficulties pale by comparison with the real problem con-
cerning microworlds in general: They are capable of performing
only because their domain is so stripped down that there is noth-
ing left that could require even the slightest glimmer of under-
standing or real perception. Perhaps the strongest testimony to
the inadequacies of the microworlds as a viable approach to com-
puter cognition comes from Winograd himself, now a professor
at Stanford, who states:

The idea is that language and thought can be modeled by such
things as formal logic. But I think that that is grossly oversimpli-
fied. What people actually do has very little in common with for-
mal logic, and what’s missing is the social dimension. Once you
take into account what you are using a word for, what part it
plays in discourse, there is no boundary to the meaning of that
word.

THE COGNITIVE ENGINE

297

Consequently, microworlds don’t appear to be the answer to the
commonsense problem. Let’s examine another line of attack.

FRAMES

Frames and scripts are predicated upon the belief that few situa-
tions we encounter in daily life are really new. Technically,
frames describe static situations, while scripts characterize a dy-
namic set of actions appropriate to a given set of circumstances.
Most circumstances that we’re called upon to deal with have
enough in common with other situations that we can distill the
principal features, analyze them, and store them for future re-
trieval and use. Thus, a frame acts something like one of those
IQ tests you encountered as a kid, where some sort of scenario is
created with a number of blanks left open to be filled in appro-
priately to demonstrate your understanding of the story. Al-
though the frame idea appears to have originated with Marvin
Minsky as an outgrowth of work on computer vision and lan-
guage, the high priest of “frameology” is Roger Schank of Yale,
a somewhat controversial character in AI circles. What
Schank’s work demonstrates is that thinking and learning are
not just passive processes of filing and retrieving information.
The mind learns to build models and structures that can be con-
tinually modified and updated as new knowledge becomes availa-
ble, and that dynamic knowledge base is used to plug the gaps in
real-life scenarios as they unfold.

As an illustration of a typical frame, here is a template for a
stock market report:

[Because of/Despite] Current newspaper headline, the market
[staged a broad advance/dropped sharply/rallied/rebounded/crept
upward/drifted lower] in [heavy/active/moderate/light] trading
with [advances/declines] leading [declines/advances] by a margin
of — to — .

Another typical Schankian example is the restaurant script,
which has slots for entry conditions like “customer is hungry”
and “table is set,” and slots for exit conditions such as “cus-
tomer has less money,” “kitchen has less food,” and “waiter has
more money.” Of course, visiting a restaurant is something we
do in stages: sit down at the table, read the menu, place the
order, eat the food, pay the bill, and so forth, so we divide the

298

PARADIGMS LOST

script into scenes. Thus, there would be an entering scene, an
eating scene, a paying scene, and so on. To allow for different
types of restaurants, the script would be divided into tracks. So
if we’re told that Alex went to McDonald’s, the fast-food track
containing the necessary variations for the way one proceeds in
such a place would be loaded. As Schank says, “We wouldn’t
place our order over a microphone at Maxim’s in Paris, nor
would we ask for a wine list at a diner.” The fact that we would
be surprised at any of these things is evidence that we have some
kind of knowledge structures containing information about what
usually happens in a given set of circumstances.

One of the acid tests for the ability of a program to “under-
stand” the situation in a given frame is for the program to be
able to answer questions about the situation, especially questions
whose answers are not directly given by the specification of the
frame. For instance, in the restaurant situation we might have
the scene:

The waitress brought the hamburger to John, but it was burned to
a crisp, so he got up and stormed out.

Now we can ask, “Did John pay for the sandwich?” On the basis
of everyday knowledge about such situations, even a small child
would have no. difficulty in realizing that John didn’t pay. But
nowhere is this rather evident fact explicitly stated. Rather it
has to be deduced from the facts that are given, together with
the background knowledge built into this particular track of the
restaurant script.

Of course, if a machine had scripts alone, it wouldn’t be able
to deal with novelty; it would understand only the prototypical
situations that had been programmed into the scripts. Conse-
quently, Schank and others, like Robert Wilensky at Berkeley,
have been busy developing programs that would know about peo-
ple’s goals and desires, and how they might go about formulat-
ing plans to achieve them. One such program was tested on the
story:

John wanted money. He got a gun and walked into a liquor store.
He told the owner he wanted money. The owner gave John the
money and John left.

Nowhere does the story make mention of robbery, nor does it
explicitly state that the gun was used to threaten the liquor

THE COGNITIVE ENGINE

299

store owner. Nevertheless, the program was able to use its store-
house of knowledge about goals and plans in order to infer these
facts.

Neither of these approaches — micro worlds and frames — has
turned out to be a panacea for the ills that plague the top-down
approach to intelligent machines; nevertheless, Simon, Newell,
Schank, & Co. continue to press on with their hopes of finally
achieving the triumph of their rule-based, “symbol-crunching”
style of AI. At this juncture, it’s well to pause to consider just
what the implications of their total and complete success would
imply for our basic question: “Can an appropriately pro-
grammed computer display strong AI, human?” In order for
top-down results to justify an unqualified yes, there would have
to be some indication of how the internal states of the machine
match up to human cognitive states, as well as to the internal
states of the brain, when they’re both performing the same sort
of task. This means that, at some level, the top-down program
states will have to make contact with actual brain states; other-
wise, the best that even a perfect top-down program could aspire
to would be weak AI, sim-human. So far the top-downers have
displayed no such points of contact, and as far as I can see no
interest in establishing such a bridge. So while it may be true
that a top-down approach can shed some light on some aspects of
human thought, it appears unlikely at present that further
pounding away at such programs is going to get us any closer to
a resolution of the basic question. Consequently, let’s move to
the other end of the telescope for a look at bottom-up attacks on
the matter of thoughts and machines.

BOTTOM-UP EMERGENCE

Herbert Simon is on record with the claim that “everything of
interest in cognition happens above the 100-millisecond level —
the time it takes you to recognize your mother.” This claim com-
pactly summarizes one of the principal axioms of faith in the
top-down school of AI: that what’s going on down at the level of
the individual neurons in the brain has no direct bearing on cog-
nition, and that somehow we can “skim off” the rules of thought
from the higher level of symbol processing and semantic net-
working, and just ignore what’s happening down below at the

300

PARADIGMS LOST

level of the microscopic processing elements. In response to
Simon’s 100-millisecond assertion, Douglas Hofstadter has writ-
ten, “I cannot imagine a remark about AI with which I could
more vehemently disagree.” Hofstadter holds to just the oppo-
site view: Everything that is important about cognition goes on
below the magic 100-millisecond level. He is one of the leaders of
the “new wave” school of AI theorists, working on the question
of how intelligent behavior can possibly emerge out of a jumble
of primitive processing elements existing at a subcognitive level.

The basic tenet of the bottom-uppers is that if we’re ever to
understand how the brain does its thing, we’re going to have to
start at the level of primitive processors functionally equivalent
to neurons, and develop theories of how cognitive states like
your mother, a 747 jet, a migraine headache, and all the other
things the top-downers attach symbolic significance to can possi-
bly come about as the result of connections and interactions be-
tween such simple processing elements.

A good analogy for understanding the bottom-up philosophy
is provided by those old-style message boards seen even today in
places like Times Square, where the news of the day and other
types of information are shown by a sequence of flashing lights
on a rectangular board. At the level of the individual lights,
there is no message: All that any of the bulbs can do is blink on
and off. However, by standing above the level of the individual
lamps themselves, we can see a properly timed sequence of such
flashing lamps as communicating the results of the World Se-
ries, a report on the state of the stock market, the outcome of an
election, or an announcement of the end of the world. The same
hardware serves for an infinite variety of symbolic messages,
but to recognize that there is a message it’s necessary, as Hof-
stadter puts it, to “jump out of the system” somehow. In some
poorly understood way, the system at the level of the light bulbs
would have to have some measure of self-awareness, or self-ref-
erence, at a higher level.

The message board example shows that whatever computation
is going on is happening not at the level of the symbolic meaning
(the message), but rather at the much lower level of the flashing
lamps. The computational rules are buried in the program that
tells each lamp when to switch on or off, not in a set of instruc-
tions for manipulating the thoughts composing the message.
This is the fundamental difference between the top-downers and

THE COGNITIVE ENGINE

301

the bottom-uppers: For top-down AI, thoughts and ideas them-
selves are passive computational entities capable of being pushed
around by the rules of a formal system; for bottom-uppers, cog-
nition involves active symbols arising from a collective of com-
putational elements at the subcognitive level — i.e., thinking is an
emergent epiphenomenon. So for bottom-up AI, subcognition at
the bottom drives cognition at the top, and the brain as hard-
ware is simply a substrate in which active symbols can interact.
It’s worth noting that this view of thinking requires some kind
of hardware in which the active symbols can interact, but there
is no absolute requirement that this material substrate be physi-
cally the same as a human brain. All that’s required is that the
playing field in which the symbols frolic have the same computa-
tional power as a human brain; i.e.; the substrate must be func-
tionally equivalent to a human brain but may differ greatly from
it in its actual physical composition.

The key element in the bottom-up program is to identify the
bridge between the “meaningless” computations at the subcogni-
tive level and the “meaningful” active symbols. One line of at-
tack has been to try to understand how we do anagrams. How is
it that given the word “ weird, ” we can immediately see that its letters
can be rearranged to form the words “ wired ” and “ wider, ” but that
no other arrangement leads to a proper word ? Surely it ’s not by
trying out all5x4x3x2xl = 120 possible arrangements
of the five letters. It seems inconceivable that the brain does ana-
grams on such a brute-force, straight-line computational basis.
Rather, we somehow use our knowledge of what letter combina-
tions tend to go together, form various groups of letters, and let
them float around in a sort of “alphabet soup” in our heads,
randomly bumping into each other and forming new combina-
tions. Those combinations that look promising are kept, while
others dissolve and drop back into the soup where they can link
up with another group. Eventually, certain combinations click
into place and a new word is formed. Hofstadter and his group
at the University of Indiana have developed a program called
Jumbo to test out various theories of how this subcognitive com-
putation (forming of letter combinations) results in the emer-
gence of active symbols (meaningful English words). The actual
strategy implemented in Jumbo provides an instructive glimpse
into the entire bottom-up program for the creation of mech-
anized thought.

302

PARADIGMS LOST

The workings of Jumbo are based upon two analogies: the way
in which a living cell conducts its chemical business, and the
fashion in which human friendships and romances are formed.
Let’s first consider the cell. The interior of a cell (its cytoplasm )
is filled with different kinds of molecules floating around in the
cytoplasmic soup. The work of the cell is carried out by enzymes,
each of which has a very specific function that it’s designed to
carry out. On each enzyme there are one or two active sites,
which will allow only a particular type of molecule to attach it-
self. The enzyme randomly wanders about in the cytoplasm until
it encounters the right type of molecule, which is then attached
to the active site. If the enzyme’s function is to join two mole-
cules (anabolic), then when the two active sites are filled, the
enzyme goes into action and joins the two molecules, whereupon
it releases the new compound into the cellular broth. Other types
of enzymes function to split compounds (catabolic reactions), or
perform more complicated functions like rearrangements and re-
groupings. Jumbo makes metaphoric use of this kind of cellular
operation by regarding the molecules in the soup as being the
letters of the given word, allowing letters to affiliate randomly
with others to form syllables, which in turn can be joined by
different sorts of enzymes to form larger groupings ultimately
resulting in proper words. In Jumbo terminology, these group-
ing operators are called codelets, and there are different types for
performing a variety of functions such as combining consonants
into clusters, consonants and vowels into syllable fragments, syl-
lables into wordlike objects, and so on. Figure 5.5 is a schematic
diagram for such an enzymatic codelet whose purpose is to em-
body the almost universal rule that, in English, the letter Q is
always followed by a U. This codelet floats around in the alpha-
betic soup until it encounters a ^-shaped character and a 17-
shaped one, each of which is captured in the appropriate half of
the codelet. Once both halves are filled, the codelet joins them
into the pair QU , thereby emptying its two slots and making
them available again to capture more characters. But once vari-
ous combinations are formed by this sort of random interaction,
how does the program decide whether or not a particular syllable
fragment, say, is promising as a step toward formation of an
actual word? This is where the analogy with human romance
comes into play.

While it may be true that the course of true love never runs
smooth, that course invariably follows a path along which recog-

THE COGNITIVE ENGINE

303

nizable landmarks appear in a time-honored sequence. First,
there is the initial contact. For boy and girl to meet, they must
come together at the same place at the same time, in order that
Nature may take its course. Following contact, sparks will start
to fly if there’s mutual interest, and the parties will begin to
explore a potential relationship by entering the next phase — dat-
ing. During the dating, each party will keep open the option of
pursuing simultaneous relationships with others, as well as the
possibility of breaking off the current budding romance. After
this period of exploration, the parties may decide to strengthen
the relationship by making it more exclusive. At this stage, while
it’s not impossible for the link to be broken by either internal
stress or external attractions, there is a deeper commitment and
it takes much stronger provocations to break it than in the ear-
lier phases. Following this courtship period, the relationship
may be even further strengthened into an engagement, which
may then be formalized socially by a marriage. Of course, de-
pending upon social conventions, religious convictions, and the
like, even such a strong bond as the marriage may ultimately be
dissolved by divorce, with the partners then being sent back into
the “social soup” to begin the process anew. Hence, every ro-
mance has to go through a sequence of increasingly tough filters,
although these steps may proceed in parallel and out of phase as
several independent relationships are being explored.

Jumbo makes use of the progressively stronger hierarchies of
bonding seen in the paths of love and friendship, in order to
decide which of the many random bonds formed between se-
quences of letters at one level should be taken seriously as candi-
dates for consideration at the next level of combination.
Consequently, along with the codelets for acting upon clusters of

304

PARADIGMS LOST

letters, the program is provided with criteria for assessing which
letter combinations seem more likely to appear in proper words.
So, for example, ee is “happier” (i.e., more stable) as a vowel
cluster than ii; nk is happier than kn, and clusters with a vowel
in the middle and a consonant on either end are happier than
those composed of three vowels. Thus senk is more likely to arise
than kniis, although when it’s discovered that senk is not a word,
it can be broken down into smaller components and tossed back
into a lower level of the soup.

At the same time the “active” enzymes are carrying out their
joining and breaking functions, another set of “passive” en-
zymes called musing codelets are imagining what would happen
if, say, one syllable was swapped for another. Without actually
making the change in the real cytoplasm, the musing codelets
consider alternative hypotheses, and explore many paths at once,
trying out various kinds of possibilities. But what is it that fi-
nally determines when all this random groping, shuffling, com-
bining, and probing finally stops? At what stage does Jumbo say
enough, and settle upon its best candidates for wordhood?

The program’s stopping rule is based upon the notion of en-
tfopy, the technical term for the measure of randomness, or dis-
order, present in the cytoplasmic soup. Initially, there are just a
lot of individual letters randomly floating around in the soup
and the entropy is high; later some structure starts to emerge as
individual letters begin to combine with others to form conso-
nant and vowel clusters, as well as short syllables, and the en-
tropy goes down; still later, syllables combine into larger
groupings and the entropy is further decreased. While all this is
going on, the enzymes are also acting to perform their specific
functions, with some enzyme operators decreasing entropy by
combining consonants into clusters, consonants and vowels into
full syllables and the like, while others, like those that inter-
change syllables within words, leave the entropy unchanged. Fi-
nally, the actions of enzymes that break bonds established by
earlier joining enzymes result in actually raising the entropy
level of the soup. Roughly speaking, the entropy level can be
thought of as a kind of “temperature” of the cytoplasmic soup,
and when the enzymes can no longer act to reduce the tempera-
ture, Jumbo stops and the clusters that remain in the soup are
taken as its best effort at forming meaningful words from the
initial alphabetic hodgepodge.

THE COGNITIVE ENGINE

305

A program to do anagrams may appear trivial to some, frivo-
lous to many, and far removed from the general idea of thinking
for almost everyone; nevertheless, Jumbo captures in particu-
larly transparent form the central ingredient in the bottom-up-
pers’ cognitive paradigm: Whatever intelligence the program
displays has not been directly programmed by specifying rules
for passive symbol manipulation. Rather, the cognitive behavior
emerges as a statistical property of many small things designed
to interact with one another that have been built directly into
the program. Consequently, in contrast to the underlying princi-
ple of top-down AI, there are no overall “rules of thought”
deterministically governing the manipulation of symbols, and
there is no central controller or manipulator and no central pro-
gram; only a vast number of individual “collectives” whose ac-
tions trigger the actions of other collectives resulting in new,
more complex patterns of organization. In short, there’s no body
doing the thinking, only a collection of somebodies.

Hofstadter & Co. have employed the same “statistical emer-
gence” principle in another program aimed at identifying letter-
forms (how do we recognize that the symbols A, A , a, and a are
all instances of the same letter?), as well as in a program for
doing analogies (ABC is to ABD as PQR is to ??). Perhaps not
surprisingly, these ideas have not caught the fancy pf the main-
stream AI community, with its historical bias dominated by rule
followers of the Simon-Newell-Schank persuasion, and expert-
system peddlers of the Feigenbaum school, who appear to be to-
tally uninterested in any type of AI not marketable in corporate
boardrooms or on Wall Street. Certainly Hofstadter’s harshest
and most vitriolic critic has been Simon’s colleague-in-arms Allen
Newell, who complained that one of Hofstadter’s papers was

“. . . somewhat polemical and diffuse with an abundance of strong
opinion and argumentation from general conceptual considera-
tions and the absence of concrete scientific data or theory to build
on. There is an abundance of attacks on the general opinions of
others, with a corresponding promotion of the general opinions of
self.

A disciple of the Schank school, Richard Granger of the Univer-
sity of California, Irvine, says:

His [Hofstadter’s] AI work is far from the mainstream. He’s a
loner. His opinions are one man’s view. . . . You have to under-

306

PARADIGMS LOST

stand that Hofstadter has recognition because he won the Pulitzer
Prize. He’s a good writer. He’s a smart, very clever person. But
that doesn’t mean that he’s right about AI.

To this kind of criticism, Hofstadter replies that

AI people seem trapped in their already-formed modes of thought
and their preconceptions. They tend to eschew the whole question
of what consciousness means. They avoid the questions of philoso-
phy of mind.

Another researcher following an evolutionary, bottom-up ap-
proach to AI is Douglas Lenat of Stanford. In his doctoral
work, Lenat developed a program called Automated Mathemati-
cian ( AM ) whose goal was to learn mathematical facts and prove
them all on its own. Lenat’s basic idea was to combine the
frame idea with evolutionary adaptation in a program that
would learn about the world of mathematical truth on its own.
Initially, the program started with a collection of frames with
slots like “Definitions,” “Examples,” and so forth. At the outset
most of these slots were empty, so Lenat provided the program
with about 250 heuristic rules of thumb that would suggest
which slot to work on next, where AM should look for new rela-
tionships between concepts, and the like. Furthermore, Lenat
provided a valuation scale by which each concept’s frame would
keep track of how each of its slots was doing by recording things
like the origin of a concept, and AM's evaluation of its worth
relative to the other frames. In this way, the valuation scheme
would act like natural selection by identifying the most interest-
ing concepts, dropping those of little “survival value.”

The results of Lenat’s work surprised even the developer.
Within a few minutes of turning on the machine, Lenat saw that
AM had discovered the concept of number. Soon after, it discov-
ered the rules of arithmetic and the idea of prime numbers.
From those building blocks of mathematics, the Fundamental
Theorem of Arithmetic (every number can be decomposed into a
product of primes in a unique way) was but a small step. Re-
grettably, after an hour or so of such set-theoretic bliss AM ran
out of steam, and began looking into such weird and self -contra-
dictory notions as numbers that are both even and odd. Upon
examining the situation, Lenat found the difficulty resided in the
heuristics that he had originally programmed in to get AM off
and running. These heuristics dealt primarily with concepts in

THE COGNITIVE ENGINE

307

set theory, and as soon as the program started departing from
this well-plowed turf, the heuristics became increasingly useless.

Learning from his experience with AM, Lenat began to de-
velop a new program, Eurisko. The basic difference between
Eurisko and AM was that Eurisko could modify not only its con-
cepts, but also its heuristics — both by the process of natural se-
lection. Lenat’s idea was to represent each heuristic by a frame
of its own. In this way, “mutations” in the heuristics could also
take place one slot at a time. The reader will recognize this pro-
cedure as strikingly reminiscent of the way Nature works in al-
tering an organism’s DNA by point mutations. The results from
Eurisko have been most publicly visible in the sound thrashing it
gave all human competitors in the national championship of the
space-war game Traveller, where the program designed space
fleets of optimal size, power, flexibility, and so forth. Lenat’s
work has been heralded by AI guru Marvin Minsky as “a whole
new field of knowledge.” Currently, Lenat is engaged in trying
to utilize the Eurisko principle to code up nothing less than the
whole field of human knowledge. He estimates that this project,
one of the most ambitious ever undertaken in the AI world, will
take at least ten years.

Right or wrong, the bottom-up movement is by now most
definitely not the work of a Lone Ranger with a couple of Tontos
riding the desolate plains of the University of Indiana’s Com-
puter Center. Variations upon the basic bottom-up theme are
popping up daily in many comers of the AI forest, and converts
are joining the fold in ever-increasing numbers. One of the most
prominent supporters is that doyen of the AI world Marvin
Minsky, who predicts that “Hofstadter is one of whom, fifty
years from now, they’ll say he was on the right track.” Minsky’s
own vision of thought, which he terms the “Society of Mind,” is
admirably captured in the Disney film Tron, in which the hero, a
hacker named Flynn, spends most of the film trapped inside a
computer, prisoner inside a system that he himself constructed.
The film shows the inside of a computer as a community of pro-
grams, each portrayed by an actor having a history, a personal-
ity, and, most important, a function within a complex political
organization. As the story unfolds, the Master Control Program
has assumed dictatorial powers, and repressive police programs
are employed to bring the other programs under central control.
Eventually, with Flynn’s help, full-scale warfare breaks out

308

PARADIGMS LOST

within the society and . . . well, I won’t spoil the ending for those
who haven’t seen the film. Rent it out at your local video shop!
Anyway, even this cursory view of Minsky’s “society,” in which
intelligence emerges from the interactions of conflicting, compet-
ing parts in a fragmented mind, shows its radical differences in
perspective from the rule-following, central actor paradigm be-
loved by the top-downers. While Hofstadter and Minsky empha-
size software in their bottom-up theorizing, let’s not forget the
hardware side of the house — the new connectionists.

The human brain is composed of about 100 billion neurons
linked together in an array of connecting axons and synapses of
a bewildering degree of complexity. Inspired by the organization
of this “wetware,” a group of computer scientists, psychologists,
and engineers have banded together to explore the hypothesis
that what produces thinking is the establishment, strength, and
reciprocal feedback of interneural connections, not computation
in the sense of manipulation of formal symbols. In short, think-
ing “emerges” from the process of neural connections’ forming
and reforming. Interestingly enough, this thesis is not a new
one: In the late 1950s, Frank Rosenblatt of Cornell produced an
artificial neural net (the perceptron ) capable of learning and
identifying a variety of letterforms. The basic structure of a
perceptron is shown in Figure 5.6, where we can clearly see the
threefold character of the machine: a lower level of individual
sensory input units wired into a higher-level array of associa-
tors (formal neurons or processors), which in turn produce the
perceived output of the device.

Unfortunately, a bit later Minsky and his MIT colleague Sey-
mour Papert showed mathematically that such a simple-minded
perceptron could never display the kinds of properties that we
would associate with genuine thinking, such as recognizing the
difference between the letters C and T. The prestige of Minsky,
Papert, and MIT, coupled with a serious misperception of just
exactly what they had actually proved, created the totally erro-
neous view that perceptronlike devices were a cognitive dead
end, resulting in a two-decade-long hiatus in development of bot-
tom-up AI in general, and connection machines in particular.
This whole sequence of events is especially ironic since, as al-
ready noted, Minsky is one of the staunchest supporters of bot-
tom-up work. Fortunately, the emergence of a new generation of

THE COGNITIVE ENGINE

309

AI workers, as well as the general shift in emphasis from serial
to parallel processing in avant-garde computer architectures,
has led to a major revival of interest in connectionism as the
royal road to machine intelligence. Let’s have a quick look at the
main planks of the connectionist platform.

The basic idea underlying connectionism is that a densely
linked network of simple, neuronlike processors can behave in
consistent ways to find certain outputs when presented with cer-
tain inputs, and that the best way to get the right outputs is not
to specify a rule for calculating them, but rather to let the sys-
tem find the right answer by trying out different connections in
the network until it settles on those that yield the correct re-
sponse. Thus, just as Jumbo engages in a directed, but still ran-
dom, assembling of word fragments into trial words and then
settles on real words by lowering the “temperature” of the lin-
guistic cytoplasm, a connection program proceeds in exactly the
same manner to identify other types of patterns such as faces,
geographic features, and letterforms. The bottom-up attitude of
the connectionists sets their program off from traditional AI in
several ways:

• Hardware counts: It’s simply not possible to separate the mes-
sage from the medium; high-level symbolic processing cannot
be abstracted from the hardware. Note that this does not
imply that all such processing, and hence thinking, must be
carried out in a medium like the human brain — only that hard-
ware constraints matter when it comes to consideration of the
cognitive powers of such processing objects.

310

PARADIGMS LOST

• Parallel architectures: Connectionist computation is carried out
in massively parallel machines. For example, Thinking Ma-
chines Corporation, a Cambridge firm founded by former stu-
dents of Minsky, who also maintains a paternal interest, has
recently began marketing The Connection Machine, a sixty-
four thousand-processor parallel machine.

• Distributed processing: Connectionist machines are deliberately
diffuse in their memory and processing; the activities are
spread around among the various processors with no single su-
pervisory controller having overall command.

• Unprogrammed: The most striking feature of connectionist ma-
chines is the relative lack of specific instructions. Rather,
there are a few general instructions, with the network finding
its own solutions by settling down into stable states instead of
following detailed, prespecified algorithms.

Currently there are several connectionist programs under
way, all employing the foregoing principles but in quite different
ways. Amusingly, one of the most active efforts is at that bastion
of top-down AI, Carnegie-Mellon, where Geoffrey Hinton and
his colleagues are building the Boltzmann machine, which is a
hardware implementation of Hofstadter’s “minimal tempera-
ture” notion, predicting the behavior of the overall system from
the statistical behavior of its parts. The Hinton group has
managed to get the machine to learn a pattern of outputs by
varying the strengths of the machine inputs. Another effort uti-
lizing the same ideas, but in a nonprobabilistic manner, is that
of Dave Rumelhart at the University of California, San Diego,
who makes each processing unit take on a range of input values
instead of being just “on” or “off.” The sum of the input values
then determines the processor’s output. Rumelhart has deliber-
ately constructed his processing elements to resemble neurons,
the signals being blurred and weighted according to which neu-
ron has transmitted them. The overall result is a system that
“relaxes” slowly into a stable state that cannot be changed by
small, random input variations. In a quite different direction,
Igor Aleksander of Imperial College in London has designed a
system that uses random samplings of an image to teach a con-
nectionist array of memory chips to respond to particular pat-
terns of inputs. One of the more striking aspects of Aleksander’s
work is that after enough inputs of “your mother’s face,” a pro-
totype of your mother’s face becomes stored in the connections

THE COGNITIVE ENGINE

311

of the machine, and thereafter the machine is able to “recog-
nize” your mother when her face again appears at the input. All
of these activities speak strongly for the importance of the con-
nectionist program as a major paradigm in the thinking machine
derby. To get a better feel for how the connectionist idea works,
let’s look at a vastly oversimplified version of a Boltzmann ma-
chine at work.

Consider the simple Boltzmann machine shown in Figure 5.7.
This machine consists of three computing elements labeled X, Y,
and Z, together with three lines of connection between them de-
noted by Wx, W2, and W,. What distinguishes a connection ma-
chine like this from a conventional computer of the type
discussed earlier is that the connecting pathways linking the in-
dividual computing elements are variable rather than fixed. This
means that each connecting link has a weight associated with it,
and this weight determines the nature of the signal that can be
passed from one computing element to another. Thus, in a con-
nection machine, it’s not just the program that dictates what the
output will be, but also the pattern of weights attached to the
links. The fact that the weights themselves are not fixed, but can
be modified during the course of the computation, enables such a
machine to display the capability for learning. Let’s see how all
this works on our simple machine in Figure 5.7.

The individual elements of the machine can be thought of as

FIGURE 5.7. A simple Boltzmann machine

312

PARADIGMS LOST

neurons in the brain, which at any moment may either be firing,
thereby outputting +1, or not firing, thus giving the output 0.
Suppose the weights attached to the links are TF, = —2, W2 =
— 1, and W 3 = + 2. These weights are applied to all firing sig-
nals transmitted along the link. So, for instance, if X and F are
both firing, X will receive a —2 input from F, and F will re-
ceive a —2 input from X. By convention, an element will fire if
and only if the sum of the signals it receives from other elements
is positive. To illustrate the way this machine works, let’s con-
struct a state diagram showing how the machine will behave
under all circumstances.

After a small amount of calculation with Figure 5.7 using the
above weights, it’s quickly seen that the machine state diagram
is as follows:

110

001 — 010

111

000

100 - 000

101

Oil - Oil

From the foregoing diagram, it’s easy to see that the machine
will always end up in one of the two stable states 000 and Oil, or
in the cycle 001 010. Since the likelihood that the machine

will end up in one of these three final states is directly propor-
tional to the number of initial states that lead to that final state,
we can say that if the initial state is chosen completely at ran-
dom, the probability of the machine’s ending up in one of the
stable states or the cycle is

P(000) = 1
P(011) = |

P(cycle) = \

Unfortunately, this example is a little too small to see the phe-
nomenon of learning, although the network has one very impor-

THE COGNITIVE ENGINE

313

tant property that can be seen: It will always go from a state of
high energy to one of lower energy. Here a high-energy state is
one in which the sum of the weights between pairs of active ele-
ments is a large negative quantity, while a low-energy state is
one in which this sum is positive. Thus we can think of the en-
ergy consumed by the network as being the force needed to oper-
ate the negatively weighted connections. It then follows that the
states that a Boltzmann machine seeks out are those of minimal
energy.

The basic principle of such a machine is that it learns to relate
certain input patterns to particular outputs. Inputs and outputs
are represented by fixing certain sets of elements by forcing
them to fire or not fire, regardless of the weights on the connect-
ing links. Thus, we could require that our elements X and Y
fire, thereby representing the input 11. We could then run a se-
ries of experiments whose goal would be to teach the machine to
output 0 whenever the input quantities were the same (00 or 11),
and output 1 whenever they differed (01 or 10). This learning
would be carried out by having a feedback mechanism by which
the machine itself successively changes the weights IT,, W2, and
W, from experiment to experiment. The basic idea is that once
the inputs and outputs have been fixed, the network is run in the
presence of random noise until it achieves a minimal-energy
state. At this point the weights attached to the connections be-
tween active elements are increased by a certain amount. Then
the inputs, but not the outputs, are again fixed and the process
is repeated, except that now when a minimal-energy state is
achieved, the weights attached to connections between active ele-
ments are decreased by the same amount as was previously
added. The result of this process is that if the second set of in-
puts sent the machine into the “right” internal state, causing it
to produce the “right” output, all the weights will have returned
to their previous value. But if the output is the “wrong” one,
some of the weights will have been permanently changed. By
this procedure, the machine will eventually come to a situation
in which most of the stable states are those that relate the
learned inputs to their corresponding outputs. Moreover, the
connections will then arrange themselves so that the network will
be able to recognize a wide range of similar, but not identical,
input patterns.

Objections to the connectionist view of cognition come in two

314

PARADIGMS LOST

flavors: theoretical and practical. On the side of theory, the big-
gest difficulty is that connectionism offers no clear-cut procedure
for getting from the low-level energy states to high-level sym-
bolic processing; i.e., there’s no prescription for bridging the gap
between computation at the hardware level and actual cognition
at the level of the software. Critics readily agree that when you
turn a connection machine on, something is likely to emerge. But
it’s not likely to be thinking. The practical objection is that in-
telligent thinking can never be done in a connectionist network
because you could never build a machine with enough connec-
tions. Connectionists reply that beyond some minimal level of
connectivity, it may be possible to substitute faster switching
speeds for more connections.

Connectionism is a very young line of research, and it should
clearly be regarded with some measure of reservation. Nonethe-
less, there is definitely something appealing about the idea of a
relatively unprogrammed machine that settles into the creation
and recognition of prototypes and patterns. Somehow this
strikes me as at least as plausible a model of thinking as a for-
mal, rule-based, highly specified machine. But in either case,
what unites the top-downers and the bottom-uppers is the con-
viction that it is indeed possible for machines to think; they are
divided only on the way the thinking is done and the way it can
be represented in a medium differing from the human brain. The
Prosecution now rests its case: Yes, machines can think! The
time has come to allow the Defense to parade its army of philoso-
phers and scientists to the stand in an attempt to convince you
that the views of the Prosecution are hopelessly and optimisti-
cally misguided. It’s to these arguments that we now turn.

PHILOSOPHERS AGAINST:

THEY’LL NEVER THINK!

Philosophers have for centuries made a questionable living out
of debating issues involving the cognitive capacity of man, and
the manner in which various facets of this capacity differ in
other forms of life. So perhaps it should come as no surprise
that the most virulent arguments heard against the idea of a
thinking machine come from the philosophers, as we have al-
ready noted in connection with John Searle’s Chinese Room ex-

THE COGNITIVE ENGINE

315

periment. It looks as if the main philosophical arguments oppos-
ing the concept of a computer’s ever having a real thought come
in three primary colors: phenomenological arguments based upon
the belief that the totality of human understanding cannot be
mechanized, logical arguments revolving around the limitations
imposed by Godel’s theorems, and antibehavioristic arguments
founded upon the notion that behavioral observation alone is not
enough to conclude the presence of genuine cognitive states. Let
me now examine each of these philosophical mainstreams in
turn.

PHENOMENOLOGY

Moses Hall on the Berkeley campus of the University of Califor-
nia is a little castlelike structure on the other side of the Cam-
panile from the massive, fortresslike concrete blockhouse of
Evans Hall, the redoubt of the Berkeley Computer Science and
Mathematics departments. This polar positioning is more than
just geographic, as over the years Moses has become the com-
mand center for a devoted band of loyalists claiming that com-
puters will never think like humans. You see, Moses houses the
Berkeley Philosophy Department, and within these hallowed
halls walks not only John Searle, he of the infamous Chinese
Room, but also Hubert Dreyfus, the philosophical bane of the
entire AI community.

Dreyfus is a small, wiry redhead with tortoiseshell glasses, a
fondness for plaid western shirts, and a burning zeal for the
existential philosophy of the inscrutable German philosopher
Martin Heidegger, who promoted the view that a rigorous expla-
nation of the mind would forever be blocked by the impossibility
of ever devising a formal representation of the whole of human
experience. Dreyfus agrees and, since such a formalization lies
at the heart of mainstream AI, he concludes that the develop-
ment of a program displaying strong AI, human, is a fool’s er-
rand; such a research effort is doomed to failure from the very
beginning. The core of Dreyfus’s claim is that many things
central to human thought, like judgment, perception, and un-
derstanding, aren’t just a matter of following rules. The mind
operates against a background of human practices, and it is this
shared social background that cannot be formalized.

In his argument against formalization, Dreyfus is joined by

316

PARADIGMS LOST

his brother Stuart, also a professor at Berkeley but in the De-
partment of Industrial Engineering and Operations Research,
who is responsible for introducing Hubert to the claims and as-
pirations of AI. As noted earlier, one of the hotbeds of early AI
work was the RAND Corporation in Santa Monica, California,
where Stuart was employed as an applied mathematician prior
to his move to Berkeley in 1967. While at RAND, he observed
the work being done by Simon, Newell, Shaw, and others on un-
derstanding, problem solving, and chess playing. While a for-
malist himself at the time, Dreyfus began having reservations
about the scientific content of what was being done in the name
of AI. During the same period, Hubert, an instructor in philoso-
phy at MIT, was hearing all sorts of wild claims from the stu-
dents of Minsky and others in the AI Lab that the philosophers
were out of date — the traditional problems of philosophy, like
perception, understanding, consciousness, and mind, were now
being solved at Technology Square on the other side of campus.
If this was indeed true, then the philosophers Dreyfus most
admired— Heidegger, Merleau-Ponty, and Husserl— must be
wrong, since one of the pillars upon which their ideas rest is the
notion that these most human of qualities cannot, even in princi-
ple, be formalized. Hubert wrote to Stuart at RAND telling him
of his MIT experiences, stating that if his philosophers were
right, then the AI work being done at RAND was barking up
the wrong tree. At this juncture, fate intervened in the form of
Paul Armer, now at Stanford but then head of RAND’s Com-
puter Science Division, who had already realized that much of
the RAND AI work was addressing deep philosophical issues,
and who felt that it would be useful to mix a philosopher or two
in with the AI crowd. As a result, and at Stuart’s suggestion,
Armer hired Hubert as a consultant to RAND for the summer
of 1964. Little did Armer realize the tempest to be unleashed by
that seemingly innocent summer consultancy.

The output of that summer exercise was a paper, “Alchemy
and Artificial Intelligence,” in which Dreyfus compared the re-
search program of AI to the attempts by medieval alchemists to
transmute lead into gold. The paper burst like a bombshell in the
AI community, being roundly and soundly denounced both as
bad philosophy and as a vicious, inaccurate attack on AI and the
motives of the AI researchers. In fact, the paper aroused such
strong emotional reactions that for several months the question

THE COGNITIVE ENGINE

317

of whether or not to issue it as an official RAND report was
debated within the upper echelons of the corporation. Eventu-
ally the matter was settled by an appeal to the principle that
just because some people didn’t like the conclusions reached by
another scholar, that was no reason to suppress publication of
the work (at RAND, anyway) if it contained no genuine logical
errors or factual mistakes. With the piece’s publication as a
RAND Working Paper (the lowest category in the RAND publi-
cation hierarchy) and the consequent implied corporate im-
primatur, the battle was joined between the AI community and
the philosophers. Ironically, the long-suppressed paper turned
out to be one of the biggest sellers in RAND publication history,
no mean feat in a list that has included such influential publica-
tions as Herman Kahn’s On Thermonuclear War, Charles Hitch’s
work on the economics of defense, and pioneering technical
monographs on game theory, computer science, and linear and
dynamic programming. Responding to this groundswell of pop-
ular support, Dreyfus later expanded the paper into the pro-
vocative book What Computers Can’t Do, exposing his
phenomenologically based objections against AI to a wider, pub-
lic audience, and has recently updated and expanded his argu-
ments in the volume Mind over Machine, coauthored with Stuart,
who somewhere along the line converted to the existentialist per-
suasion. It’s of psychological, if not intellectual, interest to have
a closer look at the style and content of the arguments that could
so uniformly incense the entire artificial intelligentsia.

The distilled essence of the Dreyfus position can be expressed
in the following syllogism:

I. The AI community claims that thinking is the
manipulation of formal symbols according to rules.

II. Phenomenology claims that knowing, understanding,
perceiving, and the like involve more than just following rules.

III. Phenomenology is correct.

THEREFORE

No amount of AI, however clever, will ever
duplicate human thinking.

It goes almost without saying that Dreyfus’s detractors question
every one of the premises on this list.

318

PARADIGMS LOST

One of the favorite arguments of the brothers Dreyfus in-
volves the way in which one acquires expertise in the perform-
ance of some task like playing chess or driving a car. In the
Dreyfus scheme of things, gaining expertise at driving a car (or
anything else) involves a successive passage through five identi-
fiable stages:

• Novice: At this lowest skill level, context-free rules for good
driving are acquired. Thus, one learns at what speed to shift
gears and at what distance it’s safe to follow another car at a
given speed. Such rules ignore context-sensitive features such
as traffic density and weather conditions.

• Advanced beginner: Through practical on-the-road experience,
the novice driver learns to recognize concrete situations that
cannot be described by an instructor in objective, context-free
terms. For instance, the advanced beginning driver learns to
use engine sounds as well as the context-free speed as a guide
for when to shift gears, and learns to distinguish the erratic
behavior of a drunk driver from the impatient actions of an
aggressive driver in a hurry.

• Competence: The competent driver begins to superimpose an
overall driving strategy upon the general rule-following be-
havior of the novice and the advanced beginner. He or she is
no longer merely following rules that permit safe and courte-
ous operation of the car, but drives with a goal in mind. To
achieve this goal, the competent driver may now follow more
closely than normal, drive faster than is allowed, or in other
ways depart from the fixed rules learned earlier.

• Proficiency: At the previous levels, all decisions were made on
the basis of deliberative, conscious choices. But the proficient
driver goes one step further and makes decisions on the basis
of a feel for the situation. There is no deliberation; things just
happen. So, for example, the proficient driver when attempt-
ing to change lanes on a busy freeway may instinctively realize
that there’s another car coming up on the blind side and delay
making a move. This instinctive reaction may arise out of ex-
perience in similar situations in the past and memories of
them, although it may appear as an unexplainable “lucky
guess” to an outside observer. Somehow there is a spontaneous
understanding or “seeing” of a plan or strategy.

• Expert: An expert driver no longer sees driving as a sequence

THE COGNITIVE ENGINE

319

of problems to solve, nor does he or she worry about the future
and devise plans. He simply becomes one with his car, and ex-
periences himself as just driving rather than as driving a car.
Thus, an expert driver has an intuitive understanding of what
to do in a given setting. He doesn’t solve problems and he
doesn’t make decisions; he just does what normally works.

The moral of this fable in five parts is that there is more to
intelligence and expertise than mere calculative rationality. Ex-
pertise doesn’t necessarily involve inference; the expert sees
what to do without applying rules. This is the essence of the
Dreyfus argument against the possibility of a rule-based pro-
gram’s ever achieving anything that even remotely approximates
genuine human intelligence.

The AI community welcomed this line of argument with about
the same level of enthusiasm as Stalin welcomed Trotsky. When
Dreyfus was invited to make a keynote address at a general com-
puter conference some years ago, the redoubtable Allen Newell
complained to the meeting’s organizers that “that kind of plat-
form gives him [Dreyfus] an authority and credibility he’s sim-
ply not entitled to.” Perhaps the most extensive critique of the
Dreyfus position was put forth by Seymour Papert, who wrote a
long reply to Dreyfus titled “The Artificial Intelligence of Hu-
bert Dreyfus.” In this lengthy document — which, interestingly
enough, was solicited by Dreyfus’s RAND sponsor, Paul
Armer — Papert accuses Dreyfus of devoting much of his argu-
ment to nothing more than gossip, with most of the remainder
composed of statements made by others that Dreyfus felt fit his
strongly held preconceptions. Other mainline AI types, like
Schank and Peigenbaum, weighed in with comments to the effect
that “everything is impossible until you do it” (Schank), and
that “every time you confront him [Dreyfus] with one more in-
telligent program, he says, ‘I never said a computer couldn’t do
that’ ” (Feigenbaum). To my mind, the most reasonable criti-
cism comes from one of the Young Turks in the field, Robert
Wilensky, who states that

certainly there are some things that are formalizable, and some
things that resist it more and more. But where do you draw the
line? And can you continue pushing the line further and further?
Those are the interesting questions, and my real objection to
Dreyfus is, why say at this stage that it’s going to fail?

320

PARADIGMS LOST

In response to his critics, Dreyfus asserts that

I would be willing to bet that in twenty years it will be settled —
that people will either be clearly on the right track or that no one
will be interested any more. And my real hunch is that in twenty
years people won’t be trying— that the wrong-headedness of their
approach will be as obvious as the wrong-headedness of alchemy.

But the arguments of Dreyfus are not the only philosophical
weapons arrayed against strong AI, human. Let’s move from
soft existentialism to hard mathematics and examine John
Lucas’s appeal to Godel’s theorems as the basis for discrediting
the very idea of a thinking machine.

MATHEMATICS AND LOGIC

Earlier we saw that Godel’s theorems show both that any reason-
ably rich formal system is incomplete and that the consistency of
such a system cannot be proved within the system itself. Fur-
thermore, in Turing’s work we saw that formal systems and ma-
chines are equivalent in what they can do. Ergo, computers are
subject to the same limitations that Godel imposed on any for-
mal system. Thus, machines are inherently limited in what they
can do and, in particular, there are statements that the mind
knows to be true but that the machine cannot prove. Interest-
ingly enough, Turing anticipated this kind of objection to AI in
his classic 1950 paper on thinking machines, replying that people
may well be subject to similar limitations. The British philoso-
pher John Lucas wasn’t convinced by Turing’s response, and
wrote a paper in 1961 titled “Minds, Machines, and Godel,” in
which he attempted to strengthen the Godelian argument against
the view that the mind is a machine or, in Marvin Minsky’s won-
derfully colorful term, a “meat machine.”

The heart of the Lucas argument takes the following course.
By standing outside the incomplete, consistent formal system,
we can see some unprovable statement to be true. But the ma-
chine cannot prove this fact; hence, a human can beat every ma-
chine, since such a true but unprovable statement exists for
every machine. Furthermore, if the human mind were nothing
more than a formal system, by Godel’s other theorem the mind
could not prove its own consistency. But humans do proclaim
their own consistency. Consequently, the mind must be more
than a machine. Since Lucas’s notorious paper appeared in 1961,

THE COGNITIVE ENGINE

321

long before any computers were being programmed to display
behavior that looked much like human thinking, most of the con-
troversy surrounding his arguments was confined to the philo-
sophical community, and provided solid testimony to support
Ludwig Boltzmann’s observation that “there is much that is ap-
propriate and correct in the writings of these philosophers.
Their remarks, when they denounce other philosophers are ap-
propriate and correct. But when it comes to their own contribu-
tions, they are usually not so.”

As with virtually all philosophical debates, the arguments
against Lucas hinge upon the precise meaning he gives to terms
like machine , as well as the hidden assumptions underpinning his
conclusions. For example, Paul Benacerraf points out that
Lucas has too limited a view of machines, since any machine that
could reprogram itself in the face of a changing environment
would be exempt from the Godel argument. Furthermore, it is
also noted that Lucas assumes that mind is consistent. In fact,
this is far from obvious, as the following paradox constructed
by C. H. Whitley shows.

Consider the sentence “Lucas cannot consistently assert this
sentence.” Lucas cannot assert the truth of this sentence even
though he can clearly see that it’s true. Why? Because if Lucas
could assert it, then that fact would undermine his assumed con-
sistency. Thus, either there is something that Lucas can see to be
true but can’t assert, or he is inconsistent. Consequently, Whit-
ley claims that Lucas holds too high a regard for humans, since
even if there is an unprovable statement that a specific machine
cannot assert, humans can’t always do it either.

Other arguments countering Lucas claim that he errs in his
application of Godel’s results. For instance, the Incompleteness
Theorem shows that a machine M cannot prove the Godel sen-
tence of M from its axioms and according to its rules of infer-
ence. But neither can mind. Furthermore, Lucas doesn’t show
that he can find a flaw in any machine, but only in any machine
that the mechanist can construct. In this same connection, it’s
well to bear in mind Godel’s own view that there could exist a
machine whose abilities equaled human mathematical intuition,
but whose program we could never understand. Nonetheless, we
would be able to set up conditions leading to the existence of
such a machine, e.g., by evolution. Thus, machines too complex
to design could nevertheless exist.

To my mind, the most intriguing arguments rely upon turning

322

PARADIGMS LOST

Lucas on his head. Rather than suggesting that our self-knowl-
edge proves we are better than machines, one could equally well
use the fact that formal systems cannot know themselves to
claim that human self-knowledge isn’t possible. In other words,
if I am a Turing machine, then my very nature forbids me to
know everything there is to know about myself. And thus does
the mathematically based argument against thinking machines
turn out to be as inconclusive as the phenomenological argu-
ments of Dreyfus. So let’s go back to the Chinese Room as a last
attempt to salvage some philosophical grist for the antimech-
anist’s mill.

ANTI BEHAVIORISM

We have already met John Searle, Dreyfus’s colleague at Berke-
ley, in connection with the Chinese Room argument. Searle is a
short, tanned, solid-looking man who speaks with a booming
voice, conveying the impression of someone born to the exercise
of power. He is also a philosopher of language of some repute,
and a virulent opponent of the syntactically based Chomskian
school of linguistics. In 1984 he was invited to give the Reith
Lectures on the BBC, an annual series in which the speaker is
charged with introducing a general audience to some of the lead-
ing intellectual issues of the day. He used this opportunity to
sharpen and extend the arguments given earlier in his Chinese
Room paper about the nature of mind and its possible connec-
tion to digital computers. Searle ’s main assertions are: (1) no
computer program is, by itself, sufficient to give a system a
mind; (2) the way the brain functions to cause mind cannot be
solely by virtue of running a computer program; (3) anything
else that causes minds would have to have causal powers at least
equivalent to those of the brain; (4) for any artifact that we
might build that had mental states equivalent to human mental
states, the implementation of a computer program would not by
itself be sufficient. Rather the artifact would have to have pow-
ers equivalent to the powers of the brain. To support these con-
tentions, Searle offers the following chain of reasoning:

• Brains cause minds.

• Minds have mental content; specifically, they have semantic
content.

• Syntax alone is not sufficient for semantics.

THE COGNITIVE ENGINE

323

• Computer programs are entirely defined by their formal, or

syntactic, structure.

Searle uses the Chinese Room to illustrate what he claims is the
commonsense, obviously unassailable nature of his position. By
now we recognize that nothing in philosophy is “obvious” and,
as would be expected, the outcries against Searle from within
the AI community are loud and long.

One of the most persistent rebuttals to the Chinese Room
is the claim that while the man inside the room may not un-
derstand Chinese, the entire system consisting of the man, the
flashcards, the dictionary-rulebook, and so on does display un-
derstanding. Searle attempts to counter this “systems” reply by
the ingenious device of internalizing the whole system; i.e., he
suggests placing the entire system within the brain of the man
by having him memorize the rulebook and the flashcards and
doing away with the irrelevant physical confines of the room it-
self. In this way, the whole system is inside the man but, Searle
argues, the man still doesn’t understand a word of Chinese. Oth-
ers base their objections on Searle’s claim that “people who ac-
cept it [the Turing Test] miss the distinction between simulation
and duplication.” For example, the philosopher Richard Rorty
argues that Searle’s insistence on the difference between the Chi-
nese Room simulation and real human thought is equivalent to
an orthodox Catholic’s argument that the Eucharist conducted
by a “demythologizing Tillichian theologian” or even an Angli-
can does not transform the wafer into the Body of Christ.

In Searle’s defense, it has been pointed out that the kind of
behavioral tests for understanding exemplified by the Imitation
Game rest on shaky empirical grounds. For example, in testing
the linguistic understanding of symbol-manipulating chimpan-
zees, Eric Lenneberg discovered that while the chimpanzees
could successfully manipulate the symbols, high school students
manipulating the same symbols (and with fewer errors) thought
they were solving puzzles, and were unable to translate a single
one of their completed “sentences” into English. Thus, just be-
cause a machine (the chimp) can successfully manipulate sym-
bols, there is no necessity for believing it understands the
language.

But not all philosophers subscribe to Searle’s pessimistic view
of the adequacy of the Turing criterion. A particularly eloquent

324

PARADIGMS LOST

defender of the Turing Test is Daniel Dennett, who stresses the
extreme generality of the test and the amount of world knowl-
edge that would be needed to pass it. Dennett concludes that
without sufficient world knowledge it would be impossible to pass
the test, and that if a machine passes the test, it’s safe to assume
that the machine has the requisite world knowledge. Thus, any
computer passing a full-blown Turing Test will be, in every in-
teresting sense, thinking. Dennett claims that Searle underesti-
mates the power of such a machine, and that it could, in
principle, actually learn to understand Chinese. In his reply,
Searle concedes this as a possibility, arguing that whether this
can actually be done is an empirical question. He then concludes
by restating his main point, that merely being an instance of a
formal system is not enough to prove that a machine is really
thinking.

After all the smoke clears away, it seems that the main contri-
bution of Vaffaire Searle to the thinking machine debate was to
provide some very much needed sharpening of the issues in-
volved in the Turing Test, as well as offering a lightning rod
upon which many conflicting views of the debate could be
focused and put on record. We now rest the case for the De-
fense: Machines will never think! But before listening to the
closing arguments, a bit of supplemental testimony from two
friends of the court will be of value in illuminating some addi-
tional facets of the overall question.

THE MORALIST AND THE MYSTIC

In the promotional material being ground out by today’s go-go
expert-system propagandists, a favored place is always reserved
for programs claiming to relieve the medical community of the
burden of deciding whether a trickle of blood at the lip and a
pain in the gut should be attributed to bleeding gums and excess
stomach acid or to a slug from a Colt .45. The hope underlying
all such programs is that by describing a bunch of symptoms,
and then correlating the symptoms with lots of rules of thumb
painfully extracted from the minds of human clinical diagnosti-
cians, it will be possible to aid physicians significantly in the
performance of their daily chores. Leaving aside the dubious
merits of such an argument, it will shed some epistemological
light on the overall question of machine intelligence if we spend

THE COGNITIVE ENGINE

325

a moment examining some of the historical accidents lending a
mild aura of credibility to the idea of a “mechanical doctor.”

THE MORALIST

Around twenty years ago, there was a brief flurry of interest in
the notion of creating computer programs that could act the role
of a psychotherapist in conversing with mentally ill patients. Jo-
seph Weizenbaum, at the time a young engineer at MIT, had
previously been exposed to the idea of computer-aided therapy
through an association with Kenneth Colby, a psychiatrist at
Stanford who had grown unhappy with conventional one-on-one
therapy and was looking toward the computer as a new tool for
probing the human psyche. Having been sensitized to the issues
of computer therapy, Weizenbaum put these ideas together with
his other interests in man-machine interactions to create ELIZA,
a program that could simulate, under limited circumstances, a
Rogerian-style therapist. A brief excerpt from ELIZA ’s most
publicized conversation gives the basic idea:

patient: Men are all alike.

DOCTOR: IN WHAT WAY1?

They’re always bugging us about something or other.

CAN YOU THINK OF A SPECIFIC EXAMPLE?

Well, my boyfriend made me come here.

YOUR BOYFRIEND MADE YOU COME HERE?

He says I’m depressed much of the time.

I AM SORRY TO HEAR YOU ARE DEPRESSED.

It’s true. I am unhappy.

DO YOU THINK COMING HERE WILL HELP YOU NOT TO BE UNHAPPY?

At first glance, this looks like a pretty impressive piece of
work. However, a little digging behind the scenes soon exposes
the whole “conversation” as an illusion created from a handful
of programming mirrors, with the program being designed to
respond to particular kinds of clues in prescribed ways. For in-
stance, to words like “father,” the program offers back, “tell
me more about your family”; statements involving “always” or
“all” result in requests for examples; after statements that don’t
fit into a predefined pattern, the program goes fishing with

326

PARADIGMS LOST

“does that have anything to do with . . . f” mentioning some
earlier topic that has been seen to be of significance to the pa-
tient because it was prefaced with “my.”

Weizenbaum thought of ELIZA as nothing more than a chal-
lenging programming exercise. Imagine his astonishment when
he found people taking the program in dead seriousness. His
own secretary insisted on closing the door to the computer termi-
nal room as she unburdened herself to the program, and people
were calling him at all hours frantically pleading for just a little
time with ELIZA to get themselves straightened out. One inter-
nationally known Russian computer scientist, interacting with a
companion program at Stanford, began unloading a whole pleth-
ora of fears about himself, his family, his career, and so forth
before an audience of embarrassed onlookers. If a person as
knowledgeable about the inner workings of the program as he
was could be enticed into making such intimate disclosures to
the machine, Weizenbaum felt there was indeed cause to take
seriously the moral implications of AI and the potential cost in
human terms of such widespread acceptance of the view that
human beings were basically just complicated machines. The re-
sult of his deliberations on these moral issues was his book Com-
puter Power and Human Reason, which appeared in the spring of
1976, ten years after ELIZA.

Just like Dreyfus’s book, Weizenbaum’s was greeted with out-
rage and polemical attacks by the AI community. The book ad-
vances the thesis that the information-processing view of man is
one aspect of a twentieth-century trend toward thinking of
human beings as means, rather than as ends, and toward consid-
ering contemporary social and human problems as being largely
amenable to quick-fix, technologically based solutions. Weizen-
baum forcefully asserts that the empirical evidence conclusively
demonstrates the falsity of the information-processing model of
humans and, even more important, that such a view is just plain
morally wrong. The critique ends with a call to the computer
science community not to promote a vision of human beings that
acts to dehumanize them further. He uses his belief that “the
computer . . . enslaves the mind that has no other metaphors and
few other resources to call upon” as support for his central
point: that thinking of people as programmed machines will in-
fluence the decisions we make about how we treat them in
today’s technically oriented world. Finally, he emphasizes that
there are domains where computers ought not to intrude,

THE COGNITIVE ENGINE

327

whether or not it’s feasible for them to do so. The above psychi-
atric situation is a prime illustration of the kind of domain Wei-
zenbaum has in mind, one in which interpersonal respect, human
understanding, and empathy are required.

Reviewers of the book found much to occupy their typewriters
with in projecting their own views of the technology-versus-hu-
manity conflict onto Weizenbaum’s cri de coeur. John McCarthy
noted that if something shouldn’t be done, then it shouldn’t be
done at all — by man or machine. He then went on to compare
Weizenbaum’s arguments to the objections posed by the Renais-
sance Church to dissecting the human body because it was the
temple of the soul. McCarthy further noted that “when moraliz-
ing is both vehement and vague, it invites authoritarian abuse
either by existing authority or by new political movements.”
One of the sharpest criticisms came from Weizenbaum’s former
associate Kenneth Colby, whose further work on computer-aided
psychiatry was the object of a particularly strong attack in the
book. Colby wrote:

Over the past four centuries the scientific community has come to
mistrust suppressions of inquiry, not only because they protect
the status quo but because so often the finger-wagging moralist
has turned out himself to be morally confused, piously self-serv-
ing, and irresponsibly blind to the consequences of his own oppres-
sive actions.

This is the kind of sound advice that in my opinion should flash
on the screen before the appearance of every politician and TY
evangelist. In light of recent history, even Jim Bakker and
Jimmy Swaggart might now agree!

Finally, a number of criticisms were directed not so much at
the book itself as at Weizenbaum’s personal motivations for
writing it. Some argued that Weizenbaum could no longer do
science and, tenure or not, at a competitive place like MIT the
pressure is always on to produce. Hence, the turn away from
doing science to becoming the conscience of the AI community.
These complaints again underscore the sociopsychological factor
at work in the genesis of what’s taken to constitute scientific
“truth.”

Outside the AI world, but still within the rather limited con-
fines of the general scientific community, the reception to Wei-
zenbaum’s brand of moralizing went down much more smoothly.
Writing in Datamation, one of the leading computer periodicals,

328

PARADIGMS LOST

the noted computer programming author Daniel McCracken af-
firmed Weizenbaum’s view that there were basic differences
between men and computers, differences that would never disap-
pear. The British computer scientist N. S. Sutherland, writing in
The Times Literary Supplement, said that “he [Weizenbaum]
raises important issues that are too often ignored. He repeatedly
and correctly insists that computers lack wisdom, but if comput-
ers are put to ill use, it is because we, not they, lack wisdom.”

But what does all this have to do with the basic question of
whether machines can think? By taking a moral stand against
AI, Weizenbaum has introduced an ingenious argument against
the idea of strong AI, human. Rather than basing his arguments
against machine cognition on technical and epistemological
grounds of the sort put forth by Searle, Lucas, and Dreyfus,
Weizenbaum advances the novel contention that even if strong
AI, human, is technically feasible, it is morally impossible! Of
course, he along with everyone else in the game agrees that we
are presently very far from anything that even smells like
strong AI, human, but then adds the moral imperative that
whether we’re close or far away is irrelevant, since the very at-
tempt to achieve genuine machine intelligence itself acts to un-
dermine our sense of humanity.

The only way Weizenbaum’s misgivings can be other than a
moot point is if the research needed to decide the matter is
allowed to run its full course, not prematurely terminated as
Weizenbaum suggests because it might lead to harmful conse-
quences. In fact, a rigorous adherence to the Weizenbaum dic-
tum would lead to the absurd situation in which research in
virtually every area of science would be stopped dead in its
tracks since any discovery could, in principle, lead to “dehuman-
ization” by showing that something we once thought of as the
preserve of humans alone is not so special after all. Rather than
pursue this essentially nonscientific line of thought, let’s turn
our attention to another kind of visionary, but this time one
whose ideas on mind and machine come down on the Prosecu-
tion’s side of the case.

THE MYSTIC

Rudy Rucker is a professional logician as well as the well-known
author of several popular books on mathematics, relativity, and
geometry, in addition to a number of offbeat science fiction nov-

THE COGNITIVE ENGINE

329

els traversing some of the same territory. If one can judge from
dust jacket photos, Rucker, with his almost shoulder-length hair
and studded leather jacket, looks as if he’d be as much at home
on a motorcycle on his way to a rock concert as in front of a
blackboard selling the Law of the Excluded Middle to a roomful
of heavy-lidded undergraduates. Having never met him, I can’t
say if this is true, but one thing is certain: When it comes to
matters of machines, minds, and souls, Rucker is a mystic of the
first rank.

In his book Infinity and the Mind, a semipopular account of
various logical paradoxes as well as some of the content and im-
plications of Godel’s work, Rucker devotes considerable space to
the question of whether mathematical logic can shed any light on
the matter of souls for robots. For all practical purposes, the
matter of machine souls and consciousness is indistinguishable
from what we have been terming strong AI, human, so Rucker’s
thoughts on the possibility of “machine dreams” are of some in-
terest.

According to Rucker, there are three possible views on the
question of human and robot souls:

• Mechanism: Neither people nor robots are anything but ma-
chines, so there is no reason why humanlike machines cannot
exist.

• Humanism: Human beings have souls but robots do not; there-
fore, no robot can ever be quite like a person.

• Mysticism: Everything participates in the Absolute, so it
should be possible for humanlike machines to exist.

The pro-AI forces have already argued the first view for us,
while the philosophers have waxed long and eloquent over the
second. Rucker supports AI, but with the bizarre twist of ap-
pealing to the mystical and mysterious Absolute. Let’s take a
longer look at how one could possibly take such a notion seri-
ously while, at the same time, adhering to the austere rigors of
mathematical logic.

The heart of Rucker’s argument for mysticism is, first of all,
to observe that the individual person consists of three separate
parts: (a) the hardware, composed of the physical body and
brain; ( b ) the software, comprising memories, skills, and, in gen-
eral, behavior; (c) the consciousness, representing the sense of
self or personal identity — in short, the soul. The key element in
Rucker’s position is to now note that it’s possible to replace or

330

PARADIGMS LOST

change any part of either the hardware or the software while
leaving (c) unaffected. We’re all familiar with changes in the
physical body, like artificial hearts, replacement limbs, and false
teeth, and no one would entertain for a moment the idea that
such changes in any way touch the soul. Rucker makes the not
entirely trivial extrapolation that, in principle, an artificial
brain could also replace the original and still leave the soul un-
affected. For the sake of argument, we’ll let this point pass. Al-
terations in (b), such as forgetting past experiences, learning new
skills, or more drastic changes like brainwashing of prisoners
of war, also happen without our ever feeling that the per-
son’s essential identity has been modified. So what remains for
part (c), the soul? According to Rucker, only the primal feeling
of existence: Descartes’s sum , “I am!” This is the only thought
that ties us to what we were in the past, or what we may become
in the future. This observation is then used to conclude that
mere existence means to have consciousness. From here it’s
smooth sailing and but a small step to the mystic’s claim that
everything participates in the Absolute, where the Absolute is
identified with existence, and hence there is no logical obstacle to
machines’ having souls (consciousness) just like human beings.
Thus, Rucker states that where the classical pro-AI materialist
would argue that “men are no better than machines,” the mystic
replies by claiming that it’s just the other way around, that
“machines can be as good as men.”

We have now run our course and, to some degree, come full
circle from formal systems as thinking machines, through philo-
sophical and moral objections to the very notion, and on to robot
souls and the universal Absolute. Finally the time has come to
listen to the closing arguments and summarize the competing po-
sitions, before retiring to the jury room to ponder the verdict.

SUMMARY ARGUMENTS

Our odyssey through the labyrinths of psychology, computer
science, mathematics, and philosophy started with the decep-
tively simple query “Can machines think?” Along the way, we
elaborated and sharpened the question of interest to the claim
that “an appropriately programmed computer can possess states
that are functionally equivalent to the cognitive states of a

THE COGNITIVE ENGINE 331

human brain,” and more succinctly expressed this assertion as
“strong AI, human.” The following diagram compactly summa-

rizes the overall situation:

Computers

Minds

Machine states

: Brain states

Programs

: Cognitive states

Briefly, the pro-AI forces (top-down or bottom-up) claim that
it’s possible to fill in the double arrows with convincing scientific
arguments, backed up with actual working computer programs;
the anti-AI community says, no way! Tables 5.1 and 5.2 show in
abbreviated form the kinds of arguments presented by both
sides to support their respective positions. After inspecting the
tables, let’s step into the jury room and come to some sort of
judgment on the question of thinking machines i

YES,

COMPUTERS CAN THINK!

PROMOTER

RESEARCH PROGRAM

(top-down school)

Turing, Dennett

Simon and Newell

Schank, Wilensky

Imitation Game

rule-based symbol manipulation
script following, frames

(bottom-up school)

Hofstadter, Lenat
Minsky

Hinton

Rumelhart

subcognitive modules
“Society of Mind”
Boltzmann machine, statistical mechanics

deterministic neuronal network

( mystical school)

Godel

Rucker

evolution of “incomprehensible” machines
universal participation in the Absolute

TABLE 5.1. Summary arguments for the Prosecution

332

PARADIGMS LOST

NO, COMPUTERS CANNOT THINK!

PROMOTER

ARGUMENT

Searle

Chinese Room

the Dreyfuses

phenomenology

Lucas

Godel’s theorems

Weizenbaum

immorality

TABLE 5.2. Summary arguments for the Defense

BRINGING IN THE VERDICT

The evidence having been submitted, the arguments heard, and
the pros and cons weighed, I vote for conviction and cast my
ballot with the Prosecution in favor of the possibility of strong
AI, human. My reasons? Well, as Sherlock Holmes so wisely
noted in The Adventure of the Beryl Coronet, “When you have
eliminated the impossible, whatever remains, however improbable,
must be the truth.” Basically, I tried to take the arguments
summarized in Tables 5.1 and 5.2 and eliminate as many con-
tenders as I could on the grounds of, if not impossibility, then at
least implausibility, or in some cases what seemed like pure so-
phistry or just plain irrelevance. Let’s examine the arguments
against the philosophers first, in increasing order of difficulty.

To my mind, Weizenbaum’s argument from morality can be
dismissed at the outset as fundamentally irrelevant to the ques-
tion to be decided, namely, whether it is in the realm of possibil-
ity for a machine to think. While I accept the position that
scientists bear some measure of responsibility for keeping tabs
on the possible social consequences of their work, and even for
making these potential consequences known to a wider audience,
the possible dehumanizing effect of a genuine thinking machine
seems to me to have no bearing whatsoever on the possibility of
actually constructing such a device. In fact, if anything I think
such potentially negative social and psychological consequences
serve as added motivation for pushing on with the research
needed to settle the matter. Either strong AI, human, is possible
or it’s not; if it isn’t, then Weizenbaum has raised a moot point;
if it is, then it’s important to know the nature and degree of that

THE COGNITIVE ENGINE

333

machine intelligence, as this is precisely the sort of knowledge
needed to decide just exactly what kind of machine we really are.
So, all things considered, I think it’s safe to eliminate Weizen-
baum’s case from the competition.

It seems almost as easy to drop the Dreyfus argument from
the list of contenders. The core of the Dreyfuses’ claim is the
phenomenological assertion that many crucial aspects of human
thinking like judgment, understanding, and perception cannot
be formalized. To support their case, the Dreyfuses present what
amounts to anecdotal evidence involving such pursuits as the ac-
quisition of skills and expertise in activities like chess playing,
driving, poetry writing, and so forth. There are many things I
don’t like about this line of reasoning, but the most important is
the ex cathedra-\ike pronouncement: Phenomenology says! On
what grounds, other than faith, can one swallow the conclusions
of the phenomenological philosophers? The whole edifice of the
Dreyfus case rests on what amounts to the religious claim that
Husserl, Heidegger, & Co. are right. But to my eye, the Drey-
fuses put forth anything but a knockdown argument supporting
this crucial assumption. Furthermore, I think it’s important to
note that they are primarily arguing against the top-down AI
programs of the Simon and Newell type. Thus, even if through
some unforeseeable set of circumstances their phenomenological
thesis could be proven correct, I fail to see how this fact would
begin to touch the program of the bottom-up school. Putting
these observations together, I think it’s also safe to scratch the
Dreyfuses from the race.

Unlike Weizenbaum’s position or the Dreyfuses’, Lucas’s ap-
peal to Godel has the surface ring of something you can really
get your teeth into: tangible, to the point, and mathematically
airtight. But Godel’s results, like all high-precision tools, apply
to a very definite and severely restricted set of circumstances,
and it seems to me that Lucas has stretched these conditions
beyond the breaking point in his efforts to invoke Godel as an
argument against thinking machines. I have already outlined
what I see as many convincing objections raised against Lucas’s
use of Godel’s theorems, so it’s not necessary to repeat them here
other than to note that Godel himself didn’t appear to see his
work as any kind of obstacle to the existence of intelligent ma-
chines. And what’s good enough for Godel is certainly good
enough for me! Thus does Lucas, too, fall by the wayside.

334

PARADIGMS LOST

Oddly enough, I find Searle’s argument based on the first-
person perspective of the Chinese Room to be the most compel-
ling, and it’s with some trepidation that I finally cast it aside
along with the others. The two axioms underpinning Searle’s
claims are (1) brains cause mental states, and (2) no amount of
syntax alone can ever generate semantics; i.e., no amount of
form will ever produce content or meaning. Personally, I have
reservations about the first point and completely disagree with
the second. To begin with, when Searle uses the word “brain” he
means the kind of human brain that each of us has sitting up
there between our ears. He later goes on to say that any pro-
gram that satisfied the conditions for strong AI, human, would
have to have the causal powers of just exactly this kind of brain.
While I definitely subscribe to the view that hardware counts, I
don’t see any compelling reason why those mysterious “causal
powers” couldn’t be present in a machine, too. As Daniel Den-
nett has put it, strong AI presupposes that “it ain’t the meat,
it’s the motion,” while Searle believes that “it’s the meat,” and,
moreover, only the human brain is the right kind of “meat.”
Without something more substantial to support this contention,
I’m afraid it’s unacceptable to me. In fairness to Searle, he has
stated that whether any entity other than the human brain could
have the right “causal powers” is really an empirical question.
So let’s grant this “meat versus motion” point to Searle and
consider his second pillar, semantics from syntax.

The real heart of the Searle case is that no amount of formal
symbol processing will ever enable a system to “understand”
what the symbols actually “mean.” Referring back to the Times
Square message board and its flashing lights, the claim is that no
matter how long and hard the board works at switching those
lights on and off, it will never “know” whether it’s telling you
about tomorrow’s weather in New York or today’s coup in
Gabon. All it knows or ever will know is the flashing of lights
according to specified rules, i.e., pure syntax. I completely dis-
agree with this claim, at least as Searle states it. The crux of our
disagreement is very simple: Searle fails to note that the “syn-
tax ^semantics” conclusion may fail when one observes the syn-
tax at a higher level. Thus, at the level of the flashing lights
there is indeed only form; however, if the message board could
somehow jump outside this level and look at itself (as we do),
then new possibilities would appear, among them the emergence

THE COGNITIVE ENGINE

335

of content from form. In order to make this kind of level jump,
the system must possess some concept of self-reference. While it’s
true that message boards are not known to contain internal mod-
els of themselves encapsulating such “self-seeing” abilities,
other kinds of systems do display such a capacity.

The canonical example of such a self-referential system is a
living cell, in which the information coded in the cellular DNA
has both syntactic and semantic content, both of which are used
in the cell’s metabolic and reproductive cycles. The point here is
that the chemical sequence on the DNA strand can, when seen at
one level, look like pure syntax, while at another level the identi-
cal chemical sequence can be interpreted, and thereby acquire
semantic content from what originally appeared to be pure syn-
tactic form. It seems unlikely that this dual-level property of the
cell has always been present, having most likely evolved over the
millennia under evolutionary pressures. So I see no reason why
the same kind of evolutionary “emergence” could not happen
with machines. In fact, this is exactly the sort of thing that
Godel seems to have had in mind when he spoke of the possibility
of our being able to set up the conditions for the coming into
existence of a machine whose program we could not understand.
Such a machine, too complex to understand, could nevertheless
evolve and be empirically discoverable. So I’m afraid that I’ll
have to also reject Searle’s claim, and with it the last, best hope
for a convincing philosophical case against strong AI, human.
Now, since I’m dreaming in print anyway, allow me to take a
moment and comment upon the various AI schools and give a
somewhat prejudicial assessment of what I see as the plausibil-
ity of their respective research programs.

As with all religions, the only thing all the AI faithful can
agree upon is the answer to the basic existential question: Is
strong AI, human, theoretically feasible? All are in accord that
the answer is definitely yes, and that we are far away from hav-
ing reached this computational state of grace. In fact, the barri-
ers separating the various believers are their different manners
of achieving salvation, that is, the philosophies they employ in
writing what they hope will be the first genuinely cognitive pro-
gram. Not to get too exercised at the outset, let’s tackle the mys-
tics first.

The mystical school is easy. Its research program consists

336

PARADIGMS LOST

merely of trying to show that strong AI, human, is not logically
impossible. In this limited but essential task, it seems to me the
mystics succeed. Unfortunately, their line of attack is a “pro-
gram without programs,” so to speak, and as a result we’re left
with the same feelings of dissatisfaction that come over us when
we encounter an indirect proof in mathematics— the kind of
proof where you assume something is true, then use that as-
sumption to derive a logical contradiction, thereby refuting the
original assumption. The most famous example is Euclid’s proof
of the infinitude of prime numbers (positive integers divisible
only by themselves and by 1). Euclid assumed there were only a
finite number of primes, and then showed this assumption leads
to a logical contradiction; hence, there are an infinite number of
primes. While the argument is logically beyond reproach, many
mathematicians (myself included) would have been happier with
a constructive proof, in which a prescription was given for actu-
ally cranking out the primes one after the other, together with
an argument showing this algorithm would never stop. Regretta-
bly, it can be shown that no such recipe for primes exists, so in
this case perhaps Euclid is as far as we can go. But when it
comes to AI, the proof is in the program, and the mystics offer
no programs. So let’s turn to the two main contenders in the AI
race, top-down and bottom-up.

At the grand scientific Academy on the island of Laputa, Gul-
liver encountered a wonderous architect who “had contrived a
new method for building houses, by beginning at the roof, and
working downward to the foundations.” As far as I can tell, the
ingenious methods of this architect seem to have been passed on
to his intellectual inheritors, the top-down Alers. Somehow the
idea of programming meaning into a set of symbols and then
letting those symbols interact according to specified rules, thus
creating a semantic network of some kind, just doesn’t have the
right feel to me. It escapes me as to why the given high-level
rules for symbol interaction need bear any natural relationship
to whatever rules the brain might actually be using — if indeed
there are any such rules, in the high-level, top-down sense of
that term. In fact, it’s manifestly clear in numerous psychologi-
cal experiments involving chess players, list memorizers, and the
like that the way the top-downers have programmed their com-
puters to perform these tasks bears little resemblance to the way
humans carry out the same activities.

THE COGNITIVE ENGINE

337

In addition, there’s the not so minor matter of human evolu-
tion. Presumably, whatever cognitive capacity the human brain
possesses was acquired somewhere along the line of development
from an earlier, protohuman, reptilian sort of brain. In other
words, the ability to represent the world symbolically and to op-
erate mentally with those symbols arose as an emergent property
out of whatever hardware happened to be available at the time.
So it seems reasonable to me to take as a working hypothesis
that there might be something special about that particular type
of hardware, and whatever that something special may be, it
cannot be omitted if you’re in the business of trying to duplicate
with another kind of hardware how humans actually think. Now
before you start thinking that this contradicts my earlier objec-
tion to Searle when he claims that probably only the human
brain is the right kind of hardware, let me hasten to note that I
firmly believe that duplication of human cognitive processes in a
machine is a feasible task. What I don’t buy is the top-down
idea that hardware doesn’t matter. In this regard, I’m totally in
sympathy with the bottom-up position that hardware is impor-
tant, but that there’s no reason to think that a brain made out of
organic neurons is the only kind of hardware that can have, in
Searle’s phrase, the right “causal powers.” I have yet to see any
convincing evidence to indicate that whatever the “something
special” is that brought about the emergence of human mental
states, it couldn’t be functionally implemented in silicon instead
of “neuron stuff.” Which brings me to consideration of the final
school, the bottom-uppers.

By now it should be patently clear that I reserve my real sym-
pathies for the thesis and program of the practitioners of bot-
tom-up AI. A crucial factor underlying my generally favorable
view of the bottom-up approach goes all the way back to one of
the foundational issues upon which the whole thinking-machine
debate rests: the distinction between a model (duplication) and a
simulation. John Searle has attached great significance in his ar-
gument to the contention that a simulation is not a duplication,
and that a machine cannot duplicate human thinking but at best
only simulate it. I’ve already dealt with what I see as the falla-
cies in Searle’s line of reasoning, but I do agree with him on the
point that simulation does not equal duplication. Since confu-
sion on this point is rampant in the AI literature on thinking
machines, this is a good moment to elaborate upon the distinc-

338

PARADIGMS LOST

tion, especially as it occupied such a central place in the formu-
lation of my views on the importance of bottom-up AI.

Suppose we have two sorts of objects, let’s say a Boeing 767
jet and a second object that someone claims is a “duplicate” or a
“model” of the 767. Just what would this mean? What would it
take to be a model of a 767? Well, it means just what any ten-
year-old kid interested in model airplanes thinks it means,
namely that there’s a direct correspondence between the external
stimuli, internal states, and behavior of the 767 and the inputs,
internal states, and outputs of the model. However, the corre-
spondence need not necessarily be either one to one or onto, so
there may be some external stimuli, states, and/or behaviors of
the 767 that are not represented in the model. So, for example,
when you go to Seattle and look at a model of a 767 in the wind
tunnel, the seats, movie screens, beverage carts, and all the other
paraphernalia forming many of the internal states of the real
767 are not present in the model, for the very good reason that
they are irrelevant to the model’s purposes, i.e., testing the aero-
dynamic properties of the real plane. Nevertheless, the external
stimuli, states, and behaviors of the model are in direct relation-
ship to a subset of the inputs, states, and behaviors of the real
plane. Such a correspondence generates a modeling relationship
between the real 767 and the object in the wind tunnel. Observe
that the model is simpler than the real thing it models, in the
sense that the model has fewer states. This property is charac-
teristic of modeling relationships: Models are always simpler
than what they model. Now what about a simulation?

In my study at home I have a brand-X laser printer whose
operating instructions assure me that by suitable fiddling I can
make it “emulate,” i.e., simulate, a different type of printer, a
Hewlett-Packard LaserJet Plus. What does it mean to say my
brand-X machine can simulate another machine? Well, it means
simply that the inputs and states of the HP machine can be
coded into the states of my machine, and those states of my ma-
chine can be decoded into the appropriate outputs that would be
generated by an actual HP printer. Note that in order for such
an encoding/decoding dictionary to be set up, my machine must
be more complicated than the HP machine in a very definite
sense. Specifically, in order for the inputs and states of the HP
machine to be encoded into the states of my “simulator,” it must
be the case that my machine has more states than the HP
printer when both are regarded as abstract machines. Thus, the

THE COGNITIVE ENGINE

339

simulator (my printer) must be more complicated than the ob-
ject being simulated (the HP printer). This situation is com-
pletely general: A simulation is always more complicated than
the system it simulates.

The foregoing brief, informal discussion of models and simu-
lations can be formalized in precise, mathematical terms, provid-
ing criteria that are, in principle, testable and that we could use
to distinguish a program that models human cognitive processes
from one that merely simulates them. In this context, it’s inter-
esting to note that a simulation of the brain would necessarily
involve a system having more states than the brain itself pos-
sesses. However, the brain, with its 100 billion or so neurons, has
at least 2'°u possible states, a number that commands some re-
spect in any company, exceeding the number of protons in the
known universe (1078) by a factor of about 2100bllllon, a number
so large it’s difficult even to write it down in words. Thus, we
can confidently predict that there will be no simulations of the
human brain in the short, intermediate, or very long-term fu-
ture. Models of the brain are another matter, and it’s fortunate
that what strong AI, human, needs is models, not simulations.

On balance, it seems to me that the thinking-machine debate is
really a battle between philosophers, regardless of the fact that
some of them may be masquerading as psychologists, computer
scientists, mathematicians, or programmers. And, as it should be
in all stories involving philosophers, the debate ends up in com-
plete chaos. My gut feeling is that a genuine machine intelli-
gence will be with us within the next decade or two, but I’ll have
to confess that that opinion is based as much upon wishing, hop-
ing, and wondering as upon hard facts and philosophical argu-
ments. But I can conclude this excursion into the world of
brains, minds, and machines with one opinion that is clear and
definite: However the matter of strong AI, human, is finally re-
solved, the outcome will radically change our view of ourselves
and our perception of the place we occupy in the cosmic order of
things.

Speaking of the cosmic order of things, the time has come to
move our consideration of the uniqueness of human beings away
from the literally mundane considerations of biochemical struc-
ture, behavior, language, and mind and into the Milky Way Gal-
axy itself, for a look at the likelihood that there are other
intelligent beings out there like us.

WHERE ARE
THEY?

CLAIM:

THERE EXIST INTELLIGENT BEINGS IN OUR
GALAXY WITH WHOM WE CAN
COMMUNICATE

THE FERMI PARADOX AND PROJECT OZMA

In a conversation with Edward Teller, Emil Konopinski, and
Herbert York at a physicists’ lunch in the summer of 1950 at
the Los Alamos labs, Enrico Fermi responded to someone’s claim
that extraterrestrial intelligences, or ETIs (be they individual
entities, group intelligences, civilizations, or whatever), exist in
our galaxy with the now-famous remark “Then where are
they?” As one might expect of a comment from Fermi, this com-
monsense question contains deep, even profound, scientific and
philosophical implications that have deservedly received much

WHERE ARE THEY?

341

scholarly attention during the intervening decades, not counting
the snowstorm of pseudoscience pulp cranked out by Erich von
Daniken and others in the “UFOs are here” genre. The pillar
upon which almost all arguments for the existence of ETI rests
is the Principle of Mediocrity, asserting that on a cosmic scale
there’s nothing special about either the Earth or human beings.
Consequently, Fermi’s question leads to the paradox that if
we’re nothing special, then intelligent life should have developed
in millions of solar systems. Yet we’ve never seen a single shred
of hard evidence to support the existence of ETI, the von Dani-
kens of the world notwithstanding. On the other hand, if ETIs
don’t exist, then we are indeed something special, in gross viola-
tion of the Principle of Mediocrity. Either of these conclusions
is mind-boggling in its implications, and steps on tender toes and
egos across the entire landscape of science. But as usual in sci-
ence, the questions, theories, and armchair philosophy vastly
outweigh the experimental evidence needed to assess them, and
it’s been only rather recently that we have finally started to ac-
quire the real data that many hopefully expect to lead to a
definitive resolution of the paradox. That story begins in 1960
with a twenty-nine-year-old astronomer named Frank Drake,
and the then rather new field of radio astronomy.

Sometime early in the morning of April 11, 1960, the 26-meter
radio telescope of the National Radio Astronomy Observatory in
Green Bank, West Virginia, was turned to the constellation of
Cetus the Whale, and Frank Drake initiated Project Ozma,
named after the princess of L. Frank Baum’s mythical land of
Oz. Drake was listening for signals from assumed intelligent be-
ings inhabiting a planetary system surrounding the star Tau
Ceti. Thus began the experimental phase to answer the corollary
of Fermi’s question and one of mankind’s oldest puzzles: Are we
alone in the universe1? Tau Ceti had been chosen as a target since
it’s not too unlike our own sun in type and age, in addition to
being “only” about 11 light-years away, a veritable next-door
neighbor on the astronomical scale of things. Drake, now a sil-
ver-haired dean at the University of California, Santa Cruz, re-
calls that when Tau Ceti disappeared over the horizon on that
first night of listening, the telescope was then turned to Epsilon
Eridani, the second target star in the experiment. To everyone’s
great astonishment, pulses at the metronomic rate of eight per
minute immediately began to pour forth from loudspeakers in

342

PARADIGMS LOST

the room connected to the telescope. The next day when Epsilon
Eridani was again visible, the pulses mysteriously disappeared,
only to reappear some days later. The second appearance, how-
ever, was also noted on a secondary antenna specifically installed
to screen Earth-based “false alarms,” thus ruling out an extra-
terrestrial origin for the pulses. Through a variety of unofficial
back-channel sources, Drake later learned that the pulses were
due to experimental military radars being tested at the time in
the relatively “clean” radio environment of the remote outback
of West Virginia. After about two hundred hours of observing
Tau Ceti and Epsilon Eridani, no legitimate signals of extrater-
restrial intelligence had been recorded, and since the telescope
was needed for other tasks, Project Ozma was brought to a close
with two definite conclusions: (1) the experimental search for
ETI (SETI) was a task well within the realm of modern tech-
nology, and (2) SETI can be dangerous to the health of radio
astronomers, with false-alarm-induced heart attacks a continu-
ing occupational hazard!

Project Ozma had actually been sparked off by a 1959 pro-
posal made by Philip Morrison of MIT and Giuseppe Cocconi of
CERX in a note to the influential British journal Nature, in
which they argued that on physical grounds the most natural
place to look for an ETI signal would be at the radio frequency
of 1420 megahertz (MHz), the frequency at which ordinary hy-
drogen, the most abundant element in the universe, naturally
radiates in the cosmic void. By happenstance, it turns out that
the background noise in outer space is very low at this fre-
quency, making the Morrison-Cocconi “waterhole” a likely place
to look for an ETI signal, at least if it’s sent by radio. Drake
immediately picked up on the proposal and saw it as a good way
both to test the newly installed 26-meter radio telescope, and at
the same time call attention to the fact that SETI had now
moved from the realm of philosophical speculation to that of ex-
perimental science.

So in the decade from Fermi’s “where are they?” to Frank
Drake s Ozma, the SETI battle lines had been drawn with the
major theoretical and experimental boundaries clearly defined:
What theoretical arguments can we give from astrophysics,
planetary science, biology, cognitive science, anthropology, lin-
guistics, and philosophy to begin to resolve Fermi’s paradox,
and what kinds of engineering, physics, and computing re-

WHERE ARE THEY?

343

sources can we bring to bear on Drake’s problem of actually de-
tecting an ETI signal? These are the scientific issues that have
dominated the SETI landscape for the past couple of decades
and around which the SETI arguments, pro and con, revolve to
this day.

THEORETICAL ETI: THE DRAKE EQUATION

Undaunted by the failure of Project Ozma to find a needle in the
cosmic haystack, Drake convened a small workshop shortly after
the conclusion of the search to examine the entire question of
ETI and to plot a course for future scientific work on the mat-
ter. As a starting point for the discussions, Drake followed the
commonly accepted reductionistic path for scientific investiga-
tions of the unknown, decomposing the ETI question into a col-
lection of individually digestible pieces involving the physical,
biological, psychological, and sociological conditions that would
have to be met for ETI to exist. The recombination of these fac-
tors led to what is now termed the Drake equation, which has
subsequently served as the starting point for almost all theoreti-
cal speculations about ETI. An understanding of this equation
is of prime importance for grasping the way in which science
has attacked the ETI question, both theoretically and experi-
mentally, so let’s take a more detailed look at Drake’s pioneering
idea.

In developing the equation expressing the number of com-
municating ETI civilizations existing in our galaxy, Drake
made the not unreasonable assumption that for us to be able to
contact such a civilization several conditions would have to be
fulfilled. These conditions can be conveniently grouped into the
following categories:

• Astrophysical and geophysical: An ETI would need to have a
suitable physical environment for development, probably on a
planet orbiting a star that, like Goldilocks’s porridge, is not
too hot and not too cold and, furthermore, not too unstable.

• Biological and psychological: It must be the case that life, as we
know it, should readily arise wherever conditions are suitable
(the Principle of Plentitude). Moreover, for the existence of
ETI we need the additional requirement that evolutionary
pressures force intelligence to emerge.

344

PARADIGMS LOST

• Sociocultural: Intelligent life must further develop into a tech-
nologically based civilization that not only persists for a suf-
ficiently long period of time but also has the desire to engage
in interstellar communication.

Clearly, satisfaction of all the above desiderata is a tall order,
and the Drake equation was developed to try to give some kind
of quantitative measure of how many such planetary civiliza-
tions might currently exist in our own Milky Way Galaxy. Now
let s look at the individual terms that by common consensus
today constitute this basic expression.

The elements forming the Drake equation are:

R* = the rate at which stars are formed in our galaxy
per year

fp = the fraction of stars, once formed, that will have a
planetary system

ne = the number of planets in each planetary system that will
have an environment suitable for life

fi = the probability that life will develop on a suitable planet

fi = the probability that life will evolve to an intelligent state

fc = the probability that intelligent life will develop a culture
capable of communication over interstellar distances

L = the time (in years) that such a culture will spend actu-
ally trying to communicate

Under the dubious (but simplifying) hypothesis that each of the
foregoing factors is independent of the others, an estimate for
the number N of advanced communicating civilizations in our
galaxy can then be made by just multiplying each of the factors
together. This yields the celebrated Drake equation for N as:

N = R * X fp X ne X f, X /, X fc X L
physical biological cultural

Thus we see that to utilize the Drake equation effectively to esti-
mate the likelihood of ETI in our galaxy requires a spectrum of
expertise that would make even a Leonardo blanch, representing
in my view one of the great multidisciplinary problems of all
time.

WHERE ARE THEY?

345

The heart of the ETI debate then comes down to the develop-
ment of scientifically defensible estimates for N. We know that
N is no less than one; some argue that N is very much larger
than one, while others claim that N is either very large or very
small. To complete the possibilities, there are those who hold to
the position that N is neither large nor small. To make sense out
of these mutually contradictory positions, it’s useful to take a
longer look at the individual pieces making up the Drake mosaic.

SLICES OF THE ETI PIE

Since the various terms in the Drake equation have been the sub-
ject of numerous book-length treatments through the years, I’ll
content myself here with giving only a highly condensed account
of some of the more important factors that need be taken into
consideration when attempting to assign actual numerical esti-
mates (guesses) to the various terms.

R* , THE GALACTIC RATE OF STAR FORMATION

Of all the terms in the Drake equation, this one is perhaps the
best understood. Theoretical and observational astrophysics over
the past few decades has succeeded in creating a picture of stel-
lar formation involving the gravitational coalescence of stars out
of interstellar galactic clouds of hydrogen, helium, ammonia,
methane, water vapor, and dust grains. As a corollary of this
work, we also have a rather detailed picture of the life histories
of stars of various masses. It turns out that something on the
order of ten stars per year are formed in the galaxy, but only a
small fraction of these are suitable candidates to support ETI.

For a particular star to generate an environment suitable for
ETI, a number of factors need to be considered. Two of the most
important are: Will the stellar environment be conducive to the
formation of a planetary system containing Earth-like planets
with liquid water, and will the star be too short-lived for life to
emerge and move along its evolutionary path to intelligence?
Current theory predicts that stars much more massive than
about 1.4 solar masses pass through their life cycles far too
quickly for living systems to emerge, while stars that are too old
would not generate conditions conducive to life, since they will

346

PARADIGMS LOST

have formed at a time before there was a sufficient abundance of
the heavy elements (iron, sulfur, calcium, and so on) currently
thought necessary for living organisms. This is because these
elements form as the by-products of supernovas, the dramatic
explosions of stars in their death throes. Fortunately, such con-
straints eliminate only about 1 percent of the stars from consid-
eration; unfortunately, there are other constraints as well.

Theoretical and observational evidence strongly suggests that
when the stellar cloud coalesces into a proto-star, the general
pattern is for the cloud to split into two more or less equal
pieces, thereby forming what’s termed a binary system consisting
of two stars orbiting each other. Numerous calculations show
that the continually shifting gravitational stresses and strains
of binary systems, not to mention the extreme temperature fluc-
tuations, create a physical environment very unlikely to support
a stable planetary system, let alone a planetary system with a
stable habitable Earth-like zone. It appears that at least half of
the stars that are not too massive and not too old belong to such
binary systems and hence must be excluded from consideration
as an abode of life. Put all these factors together with others,
including the inappropriateness of stars that are too small, as
well as stars that occupy regions too near the center of the gal-
axy where exotic events that would be fatal to most conceivable
life forms regularly occur, and the quantity R*, which started
in the region of ten stars per year, is dramatically reduced, per-
haps by a factor of several thousand. So what we need is not
just the crude rate of star formation, but the rate of formation
of stars with the “right stuff.” In astrophysical terminology,
these turn out to be what are called G-type stars like our own
sun. Consequently, when estimating R* what we’re really look-
ing for is the annual rate of formation of single G-type stars.
We’ll give specific values later, but for now the important point
is that the vast majority of stars make pretty inhospitable
homes for the kinds of organisms that we would recognize as
being alive.

fp, THE FRACTION OF STARS HAVING
A PLANETARY SYSTEM

In the process of stellar formation, a cloud of interstellar gases
begins to contract due to gravitational attraction, changing from

WHERE ARE THEY?

347

a slowly revolving amorphous blob into a rapidly spinning, pan-
cake-shaped gaseous disk. Since the rate of spin is too great for
the disk to remain stable, one of two things normally occurs:
Either the disk flies apart into a few (usually two) more or less
equal pieces, each of which then spins at a much slower rate, or
the disk throws off a small fraction ( 1 to 2 percent) of its mass
at a distance sufficiently far from the center of rotation that the
small mass has a great enough lever arm to slow down the spin
of the central disk. The reader will recognize this as the astro-
physical equivalent of spinning ice skaters who suddenly throw
out their arms to slow their rate of spin. The first case corre-
sponds to the formation of a binary (or multiple) star system of
the sort discussed above; the second represents the currently
held Anew as to how planetary systems are formed. It should be
noted, however, that these two processes may not be mutually
exclusive, since calculations indicate that a habitable planetary
system may form if the two stars of a binary system are far
enough apart, say over 20 AU (1 AU equals the average distance
between the Earth and the Sun). But conventional astronomical
wisdom dictates that planetary systems and multiples are like oil
and water: They usually don’t mix.

Our own solar system is an example of the second kind of ro-
tation-slowing process, in which about 1 percent of the original
spinning mass was thrown off in the form of the planets (most of
it in Jupiter and Saturn). During this process, though, about 99
percent of the angular momentum of the spinning cloud was
transferred to the planets (again almost all to Jupiter and Sat-
urn), leaving the central Sun with only a modest rate of spin,
low enough to preserve its stability. Since our solar system is the
only one of which we have direct observational evidence, the
question of interest for estimating fp becomes: How typical is
our own solar system? In other words, if a star does not form as
part of a multiple system, is formation of a planetary system to
be expected?

One line of attack on the planetary question is just to appeal
to the Principle of Mediocrity and say that since our corner of
the universe is nothing special, it’s likely to be the case that
planetary systems are common. Clearly, this is more of a philo-
sophical or a religious argument than a scientific one, so to move
beyond it we have two alternatives: direct observational evidence
for extrasolar planetary systems, or stronger theoretical evi-

348

PARADIGMS LOST

dence to show how the formation of planetary systems fits into
the normal process of star formation.

The difficulty with direct observation of a planet surrounding
a nearby star is graphically described by imagining a birthday
cake with a single candle placed next to the beacon atop the Eif-
fel Tower, and then trying to see the candle being blown out by
looking at it from the Postal Tower in London. In short, the
minuscule amount of light reflected by even a Jupiter-sized
planet is totally buried in the more than billion-times-greater
luminosity of the parent star. Thus at the moment the only fea-
sible method of obtaining empirical evidence for planetary sys-
tems involves searching for small irregularities in the motion of
the star due to the gravitational effects of its hypothetical invisi-
ble companions. The best candidate for such indirect detection of
a planet appears to be the star 36 Ursae Majoris A, where wob-
bles in the star’s orbit have been attributed to a Jupiter-sized
planetary companion. However, these observations have been
questioned on various grounds, and at the present time all that
can be definitely said about observations of extrasolar planets is
summed up in a remark by David Black to the 1984 Interna-
tional Astronomical Union Conference on SETI, noting that
“there is currently no observational evidence for the existence of
any planetary system other than our own.” At the time it was
expected that the Hubble Space Telescope would provide the ex-
perimental muscle needed to resolve the matter, but the tragic
Challenger accident delayed the planned 1986 launch of the tele-
scope, leaving the experimental situation pretty much un-
changed.

On the theoretical side, numerous computer simulations of the
coalescence of the interstellar gas clouds have rather strongly
suggested the likelihood of planetary systems’ emerging over a
wide range of initial conditions. Figure 6.2 is a simulation by
Stephen Dole showing the kinds of planetary systems that
emerge out of a homogeneous condensing stellar cloud of the
same mass as our solar system, when various quantities of con-
densation nuclei are injected into the cloud to provide in-
homogeneities needed to get the condensation process started.
By way of comparison, Figure 6.1 shows our solar system with
planetary distances from the Sun measured in astronomical
units (AU), while the planetary masses are given relative to the
mass of the Earth, taken to be one. Figure 6.2 shows that a vari-

WHERE ARE THEY?

349

Orbital distance, a.u.
FIGURE 6.1. The solar system

Orbital distance, a.u.

FIGURE 6.2. Hypothetical planetary systems from computer simulations

350

PARADIGMS LOST

ety of hypothetical planetary systems ultimately emerge from
such a cloud, with the different quantities of condensation nuclei
indicated by the numbers at the left edge of the figure. The ver-
tical “forks” in the figures represent the mean and the extremes
of the planetary orbits.

What’s striking about these results is the strong similarity of
the hypothetical systems to our own solar system, at least in the
sense that there appears to be a strong tendency toward the for-
mation of a planetary system consisting of a number of smaller
inner planets, together with a few outer “gas giants.” Since this
general picture persists under a wide range of random condensa-
tion nuclei, the results provide strong theoretical support to the
case for planetary systems’ being a common feature of Sun-like
stars.

The preceding discussion has focused upon planetary systems
forming during the birth process of a star. For completeness, we
might also consider the possibility of a planet existing in space
independent of a star. On physical grounds, it’s hard to imagine
how such an object could arise unless it was originally part of a
stellar planetary system and was then somehow pulled out of the
gravitational attraction of its parent star by some kind of cata-
clysmic event, e.g., a nearby supernova, or maybe a cosmic colli-
sion of some sort. In any case, it doesn’t really matter since
simple thermal equilibrium considerations make such an isolated
planet an unlikely place to find life, even if such an object does
exist. The problem is that in order for a planet to avoid getting
too hot or too cold for life to survive, it’s necessary for the
planet to radiate back into space the same amount of energy that
it receives. Unfortunately, an isolated planet doesn’t receive
nearly enough energy from the outside to support life, so what-
ever energy there is must come from internal sources. Simple
calculations show that for bodies of planetary size, the tempera-
ture gradient needed to maintain a constant 300°K (equivalent to
27°C or 80°F) at the surface is around l,000°K/kilometer, far too
hot for the planet to survive in the solid state (the comparable
figure for Earth is only 10°K/kilometer). Thus it seems safe to
eliminate such “wandering” planets as candidates for support-
ing life.

What all this adds up to is that although no planetary system
other than our own has ever been observed, the prevailing feel-
ing is that such systems are rather common around single stars,

WHERE ARE THEY?

351

and that the Hubble Space Telescope will soon confirm this prej-
udice. If so, the value of fp may soon become the best-understood
element in the Drake equation.

ne , THE NUMBER OF PLANETS HAVING AN
ENVIRONMENT SUITABLE FOR LIFE

In the SETI community it’s generally accepted that for a planet
to be a home for life there must be a plentiful supply of liquid
water. In an extremely interesting set of computer simulations,
Michael Hart showed in 1978 that if the Earth’s orbit had been
only 5 percent closer to the Sun, the primordial water vapor out-
gassed from volcanoes in the Earth’s early history would not
have condensed to form the oceans, but would have remained in
the gaseous state instead. In turn this would have prevented the
removal of carbon dioxide, resulting in a runaway “greenhouse
effect” of the sort that is now believed to have turned Venus into
a planetary version of most people’s vision of hell, with surface
temperatures hot enough to melt lead (over 800°F) and a perma-
nent cloud cover of sulfuric acid. On the other hand, had our
planetary orbit been even 1 percent greater, then the lowered
radiation from the “youthful” Sun, coupled with the reduced
greenhouse effect, would have left the Earth covered with mas-
sive glaciers. Since the surface albedo (reflectivity) of ice is
greater than that of water or land, as more and more ice formed,
more and more of the Sun’s radiation would have been reflected
back into space, the result being that the glaciers would never
melt. So it appears from Hart’s calculations that the early
Earth sailed a very narrow path between the Scylla of a Venu-
sian hell and the Chary bdis of a Martian deep freeze.

The range of orbits around a star within which a planet can
avoid both the greenhouse and glacier effects is termed the con-
tinuously habitable zone (CHZ), and varies from star to star de-
pending upon its mass. Larger stars have a bigger CHZ, but also
burn their fuel much faster, with the result that the CHZ is not
stable for the billions of years seemingly needed for evolution to
work its magic and transform the cellular slime molds into Ein-
steins and Leonardos.

Besides the CHZ, planetary size can play a significant role in
determining how suitable the planet is for life. For example,
planets much larger than the Earth will outgas more material,

352

PARADIGMS LOST

thereby enhancing the greenhouse effect. Calculations show that
if the Earth had been even 10 percent greater in mass, this effect
would have prevailed and there would have been no orbit in
which the Earth could have traveled and still retained liquid
oceans. At the other end of the scale, a planet can also be too
small to retain an atmosphere that will be effective in blocking
out the solar ultraviolet radiation that is fatal to most forms of
life. In fact, Hart also showed that if the Earth’s radius had
been even 6 percent smaller, this would have been exactly our
fate, as then the Earth’s gravitational field would not have been
strong enough to retain the ozone molecules needed to screen out
the damaging rays.

In direct contradiction to the Principle of Mediocrity, it has
also been argued that as planets go Earth is not at all typical.
The problem is that the Earth and Moon are much more like a
“double planet” system than a primary planet and a satellite.
For example, the Moon is far larger compared with the Earth
than any other satellite of a major planet in the solar system.
The large Moon has affected the Earth in many significant ways,
e.g., large ocean tides influencing the evolution of crustaceans
and amphibians, as well as the appearance of tidewater zones,
which could have helped life emerge on land. And the large Moon
is not the only thing that’s strange about the Earth.

Another anomaly is the Earth’s very strong magnetic field.
This field is much larger in proportion to the mass and angular
momentum of the rotating Earth than that of any other planet.
This magnetic field is vital for maintaining the ozone layer pro-
tecting life from deadly ultraviolet radiation. Moreover, the
Earth also has a very active, molten core. This core is responsi-
ble for all volcanoes and mountain ranges, and for the separa-
tion of continents, which, in turn, has isolated gene pools,
thereby speeding up evolution.

Recent studies claim that all these unusual characteristic of
the Earth could well be attributed to the presence of our ex-
traordinarily large Moon. It has been conjectured that the Moon
may have been “captured” in an rare encounter in which it
passed near the Earth. Since the overwhelming majority of such
encounters result in either the complete destruction or the merg-
ing of the two colliding bodies, or a simple flyby, such double
planets as the Earth and Moon are probably very rare. Thus, if
it can be shown that the presence of a large moon in a double

WHERE ARE THEY?

353

planet configuration is necessary for the emergence of life, the
term ne may indeed be vanishingly small.

Putting all these factors together suggests that just as with
stars, finding a planet with all the “right stuff” for life may
involve an extensive search, and that the quantity ne may very
well turn out to be extremely small.

fl, THE PROBABILITY THAT LIFE WILL
DEVELOP ON A HABITABLE PLANET

Since the considerations given above concerning stars and plan-
ets strongly bias our search for life to those regions of the gal-
axy bearing a strong similarity to our own, when it comes to
thinking about the likelihood of life’s emerging the most natural
approach is to consider how likely it was for life to emerge here
on Earth. Here we give only the briefest sketch of this complex
issue, referring the reader back to Chapter Two for the gory and
glorious details.

There are five basic steps through which life as we know it
today emerged on Earth:

1. Small organic molecules had to form from the
Earth’s original material.

2. These small molecules somehow had to combine
into the long chains (polymers) required for
life.

3. In some fashion the polymers had to form iso-
lated, self-reproducing systems.

4. Cells and multicellular organisms had to form
from the self-reproducing systems.

5. Evolution had to act to produce the multitude
of plant and animal species that we call life.

As noted, the first three steps on the list are what are normally
termed processes of chemical evolution, while the last two are ac-
tivities associated with biological evolution. Let’s briefly consider
how much we can say we really understand about each of these
stepping-stones to life.

According to conventional wisdom, all life on Earth is formed
out of a few organic compounds that had to have been created
from materials present at the time of the Earth’s formation.
These compounds, primarily amino acids, mononucleotides, and

Chemical

Biological

354

PARADIGMS LOST

sugars, are commonly thought to have formed out of simple ele-
ments in ready supply on the early Earth such as water, ammo-
nia, hydrogen, and methane, with energy inputs for combining
these quantities coming from lightning, volcanic heating, and ul-
traviolet rays. A famous experiment by Stanley Miller in 1953
showed that if an electrical discharge was passed through a bot-
tle full of these gases, after a week or so many organic com-
pounds would form, including amino acids. (For a diagrammatic
representation of Miller’s experiment, see page 72.)

An important aspect of these experiments is that they won’t
work at all if there is even a small amount of oxygen present. In
fact, if you try a Miller-type experiment using the present com-
position of the Earth’s atmosphere, all that results is plain old
everyday smog. Thus it’s crucial for the “primordial soup”
theorists that the atmosphere of the early Earth be highly re-
ducing (i.e., deficient in oxygen). We’ll return to this point later
as it plays a significant role in the question of intelligence.

Subsequent Miller-type experiments by other investigators
using variations of the quantities and types of gases thought to
have been present in the primordial atmosphere showed similar
results, leading to the conclusion that natural formation of the
building blocks of life out of inorganic matter seems a good bet.
Hence, the first of the five steps to life appears to be one that is
relatively easy to negotiate in any early Earth-type atmosphere.

The linking-up of the simple organic molecules into the long
polymer chains needed for life poses a bit of a problem. The sim-
ple molecules of the type formed in a Miller experiment are very
unstable, and can easily be broken apart by the same energy
sources that created them. Thus, to survive long enough even to
begin to contribute to a polymer chain, these molecules have to
be protected from solar ultraviolet rays, which were very much
stronger on the early Earth as there was no ozone layer at that
time to protect them. The obvious solution is for these molecules
to have remained in the sea, where they could easily be protected
from dissociation by lying just a few meters below the surface.
Unfortunately, when a polymer chain of such molecules comes
into the presence of water, there is a strong tendency for the
water to break the chain apart, giving us back the original prim-
itive molecules.

The foregoing picture leaves us in somewhat of a quandary: It
seems that on the one hand, the sea was necessary to protect the

WHERE ARE THEY?

355

organic compounds from ultraviolet rays, while on the other
hand, the seawater acted as a strong deterrent to the formation
of the polymers needed for life. It’s as if you were driving your
car, and every time you stepped on the gas with one foot, you hit
the brake with the other. Are there any plausible ways out of
this dilemma?

If polymerization was to take place, somehow the organic
molecules had to be isolated from water. Or failing this, we must
at least propose a mechanism by which the concentration of
these molecules could have been sharply increased in the vicinity
of the ocean. Several such possible mechanisms have been sug-
gested:

1. Evaporation of water in tidal pools

2. Partial freezing in which the water is removed as crystals

3. Volcanic heating to drive off the water

4. Attachment of the molecules to the surface of clays

Each of these processes is common and has been successfully
tested in laboratory experiments showing that polymer chains of
up to two hundred amino acids can be produced. Thus, although
this step is not quite as well understood as the way in which the
primitive molecules could have arisen, there still seems to be no
reason why polymerization could not also have come about by
fairly straightforward and common physical processes.

The last step in chemical evolution — self-reproduction — is by
far the least well understood process in the entire pathway to
life. We have already covered this step in excruciating detail in
our treatment of the origin of life in Chapter Two, so for now it
suffices to keep in mind that there are many contending routes
by which all this could have occurred, none of them especially
convincing. So for the moment we just have to say that the pro-
cesses of reproduction and replication remains a weak link in the
chain leading to life.

Once we pass into the realm of biological evolution, things
begin to get easier again. Oparin’s coacervate idea, developed in
Chapter Two, deals nicely with the question of how self-repro-
ducing polymer chains could have formed themselves into a cell,
while the well-known processes of natural selection and neo-Dar-
winian evolution provide a tested mechanism by which the many
species of plants and animals found today could have arisen over
the millennia. But this is not to say that even here there are not

356

PARADIGMS LOST

still serious questions of detail awaiting answers. For instance,
life uses only twenty different sorts of amino acids, while Miller
experiments produce far more. Why did life neglect the others?
Similarly, sugars and amino acids come in two different “fla-
vors”: right-handed and left-handed. Miller experiments pro-
duce approximately equal quantities of both, and it’s reasonable
to assume that the primordial soup contained similar propor-
tions of each type. Nevertheless, the amino acids used in living
forms are exclusively left-handed, while all sugars used are
right-handed. The only explanation we can currently offer for
this puzzling fact is that by chance the left-handed amino acids
and the right-handed sugars “took off” first, and their mirror-
image competitors were excluded by natural selection. Perhaps.
Yet the question remains open, as do a variety of others pertain-
ing to the exact manner in which life came to assume its current
form on Earth.

One final point: On Earth we find two molecular types, one
good for action (the amino acids), one good for replication and
reproduction (the nucleic acids). These two molecular types com-
pose the metabolic and genetic components of every living cell.
When it comes to the question of ETI, we can naturally ask
whether or not it would be possible to have a system in which
one molecular type does both. Essential life activity involves
preservation of the genetic information, and it’s not easy to copy
a three-dimensional object. On Earth the process is carried out
by translation from the four-letter language of the nucleic acids
to the twenty-letter alphabet of the proteins. A corollary of the
earlier questions is whether or not there are alternative alpha-
betical schemes that would do the job equally well, or even bet-
ter. Current computer experiments with “artificial life,” coupled
with advances in the informational theory of living organisms,
may offer some clues on this question of obvious relevance for
ETI. But at present we can say little more.

Taken in toto, the foregoing considerations lead to the follow-
ing conclusions: Many of the building blocks of life are almost
certain to form wherever the raw materials exist and there is a
sufficient supply of free energy. Further, many kinds of natural
processes will lead to the polymer chains needed for catalytic ac-
tivity and preservation of the genetic information. Provided
some kind of mechanism appears to get the process of self -repro-
duction going, natural selection will then take over and invari-

WHERE ARE THEY?

357

ably lead to a proliferation of life forms. Thus for assessing the
quantity ft, all seems to hinge upon the likelihood of replication
and reproduction emerging as a natural adjunct to the forma-
tion of polymer chains. At present this is completely terra incog-
nita, with opinions ranging from “It’s a one-in-a-zillion fluke
that happened only here on Earth” to “It’s inevitable wherever
life of any kind forms.”

fi, THE PROBABILITY OF THE EMERGENCE
OF INTELLIGENCE

If life is going to get into the interstellar communication game,
it’s clear that some sort of toolmaking is going to be required.
This implies intelligence of a kind. While it’s still far from clear
how necessary intelligence is for biological survival, it’s possible
to identify several steps that must be taken for the development
of a level of intelligence high enough to create the technology
needed for communication outside its own environment. These
steps include:

• Development of an atmosphere containing free oxygen

• Movement of life from the sea to land

• Emergence of hands and eyes

• Use of tools

• Appearance of social structures

Between 2| and 3| billion years ago, microscopic plankton and
blue-green algae formed and thus began the process of trans-
forming the assumed reducing atmosphere of the early Earth
into one containing large amounts of free oxygen through the
process of photosynthesis. Most organisms alive at that time per-
ished in what was for them a highly poisonous oxygen-rich at-
mosphere. But those that were able to adapt found themselves in
a position to take advantage of the increased energy available in
chemical reactions involving oxygen. Such organisms could
make more efficient use of the available food, and as a result
were able to start on the road to the development of the kind of
highly energy-intensive brain that governs what we now see as
intelligent behavior.

Photosynthesis proceeds by plants’ taking in carbon dioxide
and combining it with energy from sunlight, giving off free oxy-
gen in the process. A crucial benefit to life forms from this at-

358

PARADIGMS LOST

mospheric oxygen is that some of it forms into the molecule
ozone, which acts as an effective shield against the deadly ul-
traviolet rays from the Sun. Once this shield was in place, it was
finally safe for life to leave the sea and begin to establish itself
on land. While there’s considerable evidence to indicate that in-
telligence can emerge in sea-dwelling life forms (e.g., the ceta-
ceans), it’s difficult to imagine how the kind of technology
needed for interstellar communication could develop in an
aquatic environment.

The old saying that a picture is worth a thousand words
amply underscores the fact that our visual system is capable of
taking in an enormous amount of information at a glance. The
development of a sophisticated visual system would appear to
give a definite selective advantage in the survival game — but
only if a correspondingly sophisticated brain developed to pro-
cess the visual input. It appears that simple brains just ignore
most of this input, thus ending up with the crumbs from Na-
ture’s banquet table instead of the caviar. Hence, the emergence
of eyes acts selectively to promote a larger, more capable brain.
Ditto for the appearance of hands, which need a complex brain
to make most effective use of their inherent manipulative capa-
bility.

Hands and a complex brain make the use of tools possible.
Tools, in turn, enable us to extend the capabilities of the body in
a variety of ways, all of which contribute to freeing their posses-
sor from some of the pressures of biological evolution. As a sim-
ple example, it’s not necessary to be able to run very fast to
catch a meal if you know how to throw a rock, and you don’t
have to develop massive jaws for tearing apart your catch if you
can cut the meat up into bite-sized pieces.

The argument has been made many times that groups of
thinking animals can coordinate and plan their hunting and de-
fensive activities far more successfully than individuals acting
on their own. While a social structure does not in itself lead to
higher intellect (it hasn’t in bees and ants), it appears likely that
its adoption in animals of larger brains generally does contrib-
ute a supplementary evolutionary shove toward further brain
development.

Each of the foregoing factors contributed its share to the de-
velopment of intelligent life as we know it here on Earth. In our
quest for the kind of ETI that we could expect to be able to

WHERE ARE THEY?

359

communicate with, it’s plausible to assume that many, if not
most, of the items on our list will also appear on theirs, at least
if they want to talk with us. But on the other hand, who can
really say that what we have observed here on Earth is in any
way typical of the galaxy as a whole. Thus the likelihood of
primitive life forms’ developing intelligence remains one of the
big question marks in the Drake equation.

fc, THE PROBABILITY OF THE EMERGENCE OF A
COMMUNICATING ETI CULTURE

The development of a social structure and the capacity for lan-
guage dramatically affect the path of evolution. Prior to these
changes, evolution is primarily at the level of the individual, as
information is just passed along in the genetic shuffle. However,
as soon as a social order and language enter the scene, evolution
then proceeds to act more on the society as a whole than upon
the individual. This fact allows the knowledge of one generation
to be passed along to the next, contributing to the development
of specialized skills that can be used for the entire group. This
process is vaguely analogous to the development of multicelled
organisms from single-celled predecessors, and shows the advan-
tages of centralization and specialization in the evolutionary set-
ting. So it seems likely that once intelligent life forms, at least
some of the species will develop a social order and a technology
to go along with it. The major question that remains is: Will
they want to communicate with the stars?

Who can ever know the desires of another? The emergence of
a technologically based civilization capable of interstellar com-
munication by no means implies that it will wish to contact us.
The argument has been made that part of the age-old fascination
of humans with ETI has been our puzzlement when we look at
the stars and wonder, “What’s out there?” But suppose the
Earth were just a bit closer to the Sun and, as a result, were
continually shrouded in clouds. Would we then have any real
interest in ETI? The point here is just to emphasize the fact
that possessing an ability and possessing the desire to use it are
very different matters, and that the galaxy may well be teeming
with ETIs that are happily going their way like the proverbial
three monkeys: blind, deaf, and dumb, spending their time in
contemplation of eternal philosophical truths and deep mathe-

360

PARADIGMS LOST

matical abstractions, with no interest whatsoever in talking
with us.

In this same direction, there is also the argument that perhaps
ETIs won’t be able to talk with us even if they want to. Perhaps
their science is of such a different nature than ours that there’s
no basis for a meaningful exchange of information. Or maybe
their mathematics is based upon nonnumerical quantities that
make it impossible for us to understand what they’re doing.
We’ll return to this line of argument later, but for now we raise
it just as a possibility that, if true, would reduce the factor fc to
a negligible level.

L, THE LIFETIME OF A COMMUNICATING
CIVILIZATION

Assume there is a technical civilization out there that is val-
iantly trying to communicate. How long will it be able to perse-
vere in its efforts'? This question forms the biggest uncertainty
of all in the Drake equation, and this uncertainty is unlikely to
be reducible by any sort of terrestrial experiments. To see why,
it’s helpful to examine the many reasons a communicating civili-
zation may cease its attempts. Some of the most obvious are:

nuclear war genetic deterioration

overpopulation overstabilization

exhaustion of resources loss of interest
pollution

We have been so bombarded by the media and the doomsayers
about the dangers of the items in the first column of the list that
by now most of us probably wish to just bury our heads in the
sand, ostrichlike, and wait for these possibilities to disappear.
Since I have nothing to say on these matters beyond what’s al-
ready been well chronicled, I’ll indulge the reader’s desire and
move on to the far less familiar possibilities in the second column.

The advent of modern technology and medicine has for the
first time opened up the possibility of circumventing Nature’s
way of weeding out misfits by natural selection. Modern medi-
cine now allows not only the fittest to survive, but almost every-
one else, too. In the past, weak, sickly, or genetically defective
people had a tendency to disappear from the gene pool early in
the game. But no more. Just as those with bad eyesight no Ion-

WHERE ARE THEY?

361

ger have to worry about mistaking a hungry saber-toothed tiger
for a housecat, those with various genetic defects such as
Down’s syndrome, sickle-cell anemia, and hemophilia can now
survive and even pass these defects along in the gene pool.

Gene-splicing technology has now reached the level where some
of the deleterious effects of the foregoing genetic-programming
“bugs” can at least theoretically be weeded out of the system.
However, these techniques are themselves not without their dark
side, opening up the possibility of creating entirely new types of
human beings according to any desired set of specifications.
Who is to say what kinds of humans should be produced? In
Aldous Huxley’s Brave New World , the kind of static society
that emerged from such genetic engineering turned out to be one
totally lacking in creativity, a situation good for the government
but questionable as a basis upon which SETI is likely to be con-
tinued. So genetic deterioration is a very real threat to those
societies lucky enough to survive annihilation by nuclear war as
well as the other apocalyptic “horsemen” on our list.

Population expansion, excessive energy consumption, and the
like cannot continue forever. One possibility for the stabilization
of such processes would be for the nations of the world to agree
to halt the growth of their economies — the no-growth option.
But there are dangers in the no-growth strategy, as the forcible
elimination of economic growth may also result in an overly
static society in which scientific progress and intellectual curios-
ity have been destroyed. Economic growth and the growth of
scientific knowledge have traditionally gone hand in hand, and
suppression of one could cause the elimination of the other as
well. The no-growth option would almost certainly result in the
cessation of any sort of SETI or space exploration, perhaps even -
creating a very xenophobic society that has slid back into a
primitive, pretechnological life-style from which there can be no
hope of recovery.

Finally we come to the possibility that a communicating soci-
ety will just get tired of trying to make contact and give up.
There must be some limit to how long a civilization will try to
communicate, and the probability that it will continue to send
signals or even listen for thousands or millions of years with no
return signal is surely zero. Of course, the communicating phase
may recur in cycles, with periods of intense interest followed by
loss of interest for long periods, after which communication

362

PARADIGMS LOST

starts up again. But it’s hard to argue that the communicating
periods taken in total would necessarily be longer than the peri-
ods of silence unless some results were obtained. So from the
standpoint of communication, indifference and malaise are just
as serious threats as any of the other, more cataclysmic possibili-
ties on our list.

Here it has been possible to touch upon only a few of the more
important matters surrounding each term in the Drake equa-
tion. For detailed accounts, I refer the reader to the excellent
volumes cited in “To Dig Deeper” and now pass on to the matter
of assigning actual numbers to the terms in a valiant, but proba-
bly foolhardy, attempt to divine on theoretical grounds how
likely it is that N is greater than the magic numero uno.

ANTHROPOMORPHISMS, CHAUVINISMS,
AND ETI NUMEROLOGY

Before attaching some numerical “guesstimates” to the terms of
the Drake equation, let’s discuss for a moment a few of the bla-
tant prejudices built in to the remarks given above regarding
these terms. All of these biases trace their origin to one root
cause: our interest in ETIs of the type that we could reasonably
expect to be able not only to recognize, but also enter into some
sort of sensible communication with. A good example of the kind
of ETI that we’re not talking about here is provided in Stanis-
law Lem’s classic novella Solaris, in which the central role is
played by a sentient ocean that has been under study for years
by scientists who are able to recognize that the ocean is intelli-
gent, but who are totally unable to enter into any kind of mean-
ingful dialogue or interaction with it. Another example of this
kind is given in Fred Hoyle’s classic The Black Cloud, involving
an intelligent entity composed of a cloud of interstellar particles.
While the heavens may indeed be composed of more things than
are dreamed of in our philosophies, those philosophies are ex-
actly what determine the kinds of entities that we can and want
to interact with, and hence induce the anthropomorphic slant to
our consideration of the Drake equation. But to be explicit
about it, let’s look briefly at a few of the more important “hu-
manistic” prejudices introduced into the equation.

WHERE ARE THEY?

363

• Carbon chauvinism: A sine qua non of the kind of ETI we’re
interested in is that it be a life form capable of reproduction.
This means that there must be some chemical structure that
contains the genetic information to be passed along to prog-
eny. For any reasonably complex life form, the amount of in-
formation to be passed along amounts to millions of bits, thus
requiring the kinds of long polymer chains we have considered
above. According to the known laws of chemistry, there are
only two elements capable of forming the kind of long chain
needed: carbon and silicon. Terrestrial life is based upon car-
bon for the simple reason that, at normal Earth temperatures,
silicon is not capable of forming these chains. Only at temper-
atures below — 200°C do the chemical properties of silicon
allow it to link up into chains of sufficient length to store the
needed genetic information. Thus silicon-based life forms may
well exist, but only on planets whose oceans are filled with liq-
uid nitrogen! Unfortunately, at such temperatures chemical
reactions proceed extremely slowly (that’s why we put things
into refrigerators to slow down their decomposition), and it
seems unlikely that any such silicon-based organism would
possess a metabolic rate fast enough to generate a technologi-
cal base sufficiently advanced to enter into interstellar commu-
nication. Hence our anthropomorphic bias toward carbon.

• Star-type chauvinism: To be consistent with other an-
thropomorphic assumptions about the origin of life and the
time scale for evolution, it’s necessary to assume that a com-
municating type of ETI will be found on a planet orbiting a
G-type star like our own sun. An entertaining and scientifi-
cally plausible alternative is provided by the cheela, the main
actors in Robert L. Forward’s novel The Dragon’s Egg, which
are microscopic beings inhabiting the surface of a neutron
star. The story describes how in such an environment beings
live out their lives on a time scale millions of times faster than
ours, and indicates how it might still be possible for meaning-
ful communication to take place. However, scientifically plau-
sible speculations and the way a prudent man would bet are
two different things, so we prefer to look at G-type stars until
there are compelling reasons to do otherwise.

• Planetary bias: Our discussion of the origin of life assumes
that it arose on the surface of a planet through natural chemi-
cal processes. In other words, it was not imported from inter-

364

PARADIGMS LOST

stellar space, and it did not come about as a bolt out of the
blue from “elsewhere.” Beginning with the Swedish chemist
Arrhenius and continuing to the present day with the works of
Hoyle, Wickramasinghe, Crick, and others, fanciful proposals
have been made that life forms originated elsewhere and were
somehow transported to Earth. These are basically untestable
and therefore irrefutable hypotheses; nevertheless, strong ar-
guments can be mustered against them on purely physical
grounds. So an application of Ockham’s razor leads us to
planetary chauvinism in the absence of firm contraindications.

The above collection of chauvinisms could be greatly extended,
but I think this short list gives the general idea, namely that
there is a tremendous amount of subjectivity involved in assess-
ing the terms in the Drake equation and, as a result, any numer-
ical estimates that emerge have to be taken with several shakers
full of salt. Now let’s finally turn to the process of putting some
numbers into the equation in an attempt at least to get a feel for
the range of possibilities for the quantity N, the number of com-
municating ETIs in our galaxy.

Beginning with perhaps the first widely circulated popular
account of the SETI question, the still-influential volume Intelli-
gent Life in the Universe by the well-known Russian astrophysi-
cist I. S. Shklovskii and the Cosmos man, Carl Sagan, a number
of authors have thrown their hats into the ring and taken a stab
at estimating N numerically. Table 6.1 gives a fairly representa-
tive account of these efforts, where H represents the number as-
sociated with an optimistic scenario in which everything works
out to favor ETI, M denotes a conservative estimate represent-
ing the best guess on the basis of current scientific knowledge,
and L is the pessimistic, Murphy’s Law scenario in which Na-
ture has stacked the deck against ETI.

What kind of sense can we make of an estimate of N that
ranges all the way from N = nil (“we’re alone”) to N = at least
100 million (“the galaxy is crawling with communicating
ETI”)? Or, put another way, does the Drake equation in any
way help us in deciding whether or not it’s a good scientific bet
to invest our time, money, and energy in looking for signs of
intelligent extraterrres trial life? Some have argued that our
high levels of ignorance about most of the terms in the equation
make it totally useless as a tool for studying the ETI question;

WHERE ARE THEY?

365

TERM

SHKLOVSKII
AND SAGAN
(1966)

H M L

HART

(1980)

ROOD AND

TREFIL

(1982)

H M L

R *

10 — *

0.15

0.05

0.005

1 — *

0.5

0.2

0.025

0.30

0.10

nit

1 — *

0.1

0.001

0.20

0.05

nil

1 — *

0.1

io-20

0.50

0.01

nil

0.10 — ■ *

0.5

0.1

0.10

0.50

nil

0.10 — *

0.5

0.1

0.25

nil

> 10*

10’ 100

10fi

104

100

106

104

100

> 10*

10* 100

25 X 106 100

nil

4500

~io-:l

nil

*No upper or lower estimates given.

TABLE 6.1. Estimates for N using the Drake equation

others point out that even if the numbers are only guesses, at-
tempting to pin down numerical values for the various terms
helps us at least identify and focus our efforts on those compo-
nents of N that we know least about.

Since the job of the statistician is to attempt to provide esti-
mates for various quantities on the basis of incomplete or
“noisy” measurements, it’s of interest to consider what standard
methods of probability and statistics have to say about estimates
of N generated from the highly uncertain guesses for its compo-
nents displayed in Table 6.1. There are two points worthy of
note in this connection: First of all, the argument is often ad-
vanced that you can’t make any statements about how likely
something is on the basis of just one observation. If this were
indeed true, then approaches to estimating N from the Drake
equation would definitely be in trouble, since we have only a sin-
gle example upon which to estimate all the biological and cul-
tural terms. Fortunately it’s not true that a single observation
gives no useful information. In fact, any statistician will tell
you that a single measurement is all you need in order to esti-
mate the average , or mean, of a collection of data. Consequently,
in the absence of additional data, the best estimate you can make
of what the population of data is like is to guess that its average
is just equal to that single value you have measured. The reader

366

PARADIGMS LOST

will recognize this fact as the statistical muscle underpinning
our earlier Principle of Mediocrity: What’s happening here on
Earth is nothing special; as galactic civilizations go, we’re very
ordinary and typical.

Extending the above argument using far more sophisticated
statistical tools, Peter Sturrock has calculated the statistical
spread in the value of JV using estimates of the component quan-
tities similar to those of Table 6.1. His conclusion is that with 70
percent confidence we can say that N is between 10,000 and 100
million, while with 95 percent assurance we can fix N between
100 and 10 billion. With such enormous levels of uncertainty,
we’re not really helped very much by the Drake equation itself
in estimating N. But the analysis carried out by Sturrock shows
that around 80 percent of the dispersion comes from the high
level of uncertainty in the quantity L, the lifetime of a com-
municating civilization, and that almost half of the remaining
spread is attributable to the term fc, the likelihood that a com-
municating technical civilization will emerge. Thus, even on the
basis of a single observation, it’s still possible to employ stan-
dard statistical methodology to squeeze useful information out
of the Drake equation.

A second statistical point to consider is that when we use the
term “probability” in regard to the Drake equation, we’re not
using it in the same sense as when you say that the “probabil-
ity” of the toss of a fair coin resulting in heads is \. In this
more conventional usage, the value Prob (Heads) = \ is derived
by repeating the experiment many times and then observing that
in the long run, the event heads comes up half the time. This is
the so-called relative frequency approach to estimating the proba-
bility of an event. With the exception of the astrophysical terms,
we have only a single experiment upon which to base our esti-
mates of the terms in the Drake equation. Thus, when we speak
of the “probability” of the emergence of life, or the “probabil-
ity” of the development of a communicating civilization, we are
clearly using a different kind of probability than in the coin-
tossing situation. Probabilists and statisticians call this kind of
probability a subjective probability, since its numerical value is
determined not by repeated experiments, but rather by the expe-
rience, judgment, and gut feeling of the investigator. While
such estimates are less precise than conventional probabilities
calculated by the relative frequency approach, they are not to-

WHERE ARE THEY?

367

tally arbitrary either, since various internal-consistency condi-
tions relating different estimates have to be obeyed. These sub-
jective estimates are bound to improve as we carry out further
laboratory experiments into the origin of life, as well as into our
linguistic and cognitive capabilities, and as we continue to pur-
sue investigations into the ways and means by which our collec-
tion of terrestrial cultures can avoid destroying themselves.

Freeman Dyson is a slender, dark-haired, youthful-looking
man of average height, with a long hawklike nose and the in-
tense, penetrating look of someone dedicated to his work. In
pursuit of that work he has become one of America’s premier
theoretical physicists, as well as a thinker deeply concerned
about the long-term prospects of a world in which there is
enough nuclear weaponry to provide the explosive power of a
ball of dynamite six feet in diameter for every man, woman, and
child alive on the planet today. From his intellectual redoubt at
the Institute for Advanced Study in Princeton, New Jersey,
Dyson has through the years tossed out regular tidbits of fact
and speculation creating major waves in the small pond of
SETI.

At a joint U.S.-U.S.S.R. meeting to discuss SETI held in
1971 at the Byurakan Astrophysical Observatory in Soviet Ar-
menia, Dyson made the characteristically provocative remark,
“To hell with philosophy. I came here to learn about observa-
tions and instruments and I hope we will soon begin to discuss
these concrete questions.” Thus did he succinctly highlight the
point that despite the utility of the Drake equation as a theoreti-
cal basis for many fascinating speculations about ETI, in the
final analysis it is not armchair speculation but nuts-and-bolts
experimentation that will ultimately settle the issue of whether
N = 1 or N > 1. As an amusing aside, Dyson notes in a later
account of the Byurakan meeting how he almost succeeded in
creating a minor diplomatic incident when he wrote what he
thought was the Russian translation of the English word “phi-
losophy” on the blackboard as part of the opening sentence of
his call to arms. It seems that at least in 1971, the Russian word
filosofiya was used in a very specific sense to denote the kind of
Marxist “philosophy” forming the basis for the Soviet political
ideology, and not philosophy in the more general sense under-
stood in the West. Fortunately Dyson consulted a Russian

368

PARADIGMS LOST

friend on the translation of the remainder of his opening sen-
tence before it went on the board, thereby averting an awkward
moment, but also showing the delicacy needed to communicate
even with earthly intelligences. Anyway, taking our cue from
Dyson, let’s now turn away from theoretical SETI and pay some
attention to the problem that started the SETI ball rolling in the
first place: listening for radio signals from ETI.

EXPERIMENTAL SETI:

HOW SHOULD WE LISTEN?

Imagine the following situation: You’re an American who has,
for reasons unclear even to yourself, taken up residence in a
small Central European country. On balance, you’re not too
sorry to have left behind most of the dubious delights of Ameri-
can culture: neighborhood junk-food emporiums, carbon-copy
shopping malls, and the clownish preoccupations with cars and
cholesterol, “relationships” and real estate. Nevertheless not all
of your cultural baggage has been discarded, and your heart still
beats a little faster when the shadows start to lengthen and the
football stadiums from Stanford to Yale begin to fill. Unfortu-
nately, you don’t, live within broadcast range of the U.S. mili-
tary TV stations in Europe, so it’s not possible to tune in to
your favorite distraction nor will you again experience the bit-
tersweet pleasures of those regular autumn meetings with your
bookie. However, your spirits immediately perk up when a
friend calls from America with the welcome news that one of the
cable TV companies is putting up a new satellite that will regu-
larly transmit all the football feeds from every network, major
and minor, directly to a variety of sister stations all over
Europe. To tune in to this bonanza, all you need do is crank up
your parabolic antenna and settle back for an autumn’s worth of
the life you always aspired to — a steady dose of American foot-
ball without having to actually be there.

Unfortunately, your friend is not exactly the technical type,
leaving you in the dark as to how, when, and where to point your
antenna to start harvesting this bounty of flying footballs and
petulant player strikes. So what kinds of difficulties do you have
to overcome? First of all, there’s no information about how
strong the signal from the satellite will be, so you don’t know

WHERE ARE THEY?

369

how sensitive your antenna must be. To be on the safe side, you
buy the biggest dish your landlord will allow on the roof. Next,
you have no information about the frequency (station) on which
the satellite will be broadcasting, so to cover all bets you buy a
receiver that will scan all channels. Moreover, the cable com-
pany’s signal may not be perfectly pure, so you need to have a
fairly broad-band receiver allowing you to pick up Channel 4
even if the signal that’s coming to you is really Channel 4.2 or
Channel 3.8, say. Furthermore, there’s no information about the
satellite’s orbit or broadcasting schedule, so you have to play
guessing games with the company as to exactly where in the sky
and when you should point your antenna to try to pirate the
signal. Finally, even should you manage to surmount all of these
hurdles and actually tune in to the broadcast, you’ll find that the
cable company engineers are no dopes, and that before visions of
the Rose Bowl will appear on your screen you’ll be faced with
the problem of trying to decode the signal. Now let’s add a bit
more spice to this stew by recalling that your friend (like most
of mine), well meaning as he is, blows at least as much smoke as
fire, so there may not even be a satellite! So now what do you
think of your chances of catching Notre Dame doing battle
against USC on the tube this fall?

This sad little story is but a pale imitation of the difficulties
facing the experimental seekers of ETI. As it’s commonly ex-
pressed in SETI circles, it’s like a blind man in a dark room
looking for a black cat — a cat that might not even be there! To
get some feel for the real magnitude of the problem, let’s take a
little harder look at the three most important factors in a radio
search for ETI:

FREQUENCY

The Earth’s environment is filled with all sorts of radio noise
coming from sources ranging from TY stations and military ra-
dars to various geophysical activities going on beneath and upon
the planet’s surface. This noise tends to drown out the reception
of a certain band of frequencies from outer space. But outer
space itself is far from quiet, containing its own brand of radio
noise stemming from cosmic events, not to mention the constant
background radiation from the original Big Bang. Figure 6.3
shows how these two kinds of radio noise combine at the Earth’s

370

PARADIGMS LOST

surface to screen out effectively a wide spectrum of radio fre-
quencies.

As noted earlier, every molecule acts as a miniature radio
transmitter radiating at its own characteristic frequency. In the
figure, the frequencies for interstellar hydrogen (H) and the hy-
droxyl radical (OH) are marked, clearly showing their favored
positions near the rather flat minimum on the thermal noise
curve. It was for this reason that Morrison and Cocconi pro-
posed looking for ETI signals at a frequency near 1420 MHz.
The region between hydrogen and the hydroxyl radical has, for
obvious reasons, come to be termed the waterhole in SETI cir-
cles, reflecting not only the chemical composition of water
(H20), but also the metaphorical interpretation of a waterhole
as a meeting place for all sorts of “animals.”

At present most radio searches are being conducted in or near
the waterhole frequencies, although there are occasional propos-
als to look at other frequencies when seeking special types of
signals. But so far there have been no convincing reasons offered
to depart from the basic arguments laid down by Morrison and
Cocconi, and it seems reasonable to suppose that the majority of
Earth-based searches will continue to stay in this region. We’ll
return to this point in more detail later in the chapter.

WHERE ARE THEY?

371

SENSITIVITY

On the basis of economy, it’s a good bet that the kind of signal
an ETI would transmit will come in at least two parts: a beacon
to attract our attention, as well as a second signal containing the
information to be conveyed. These two types of signals have
vastly different frequency requirements. To attract attention
over the largest possible distance, all the power in the trans-
mitter needs to be channeled into a single wavelength forming a
beacon that will stand out against the cosmic background. On
the other hand, when you’re transmitting information, the wider
the range of frequencies you can transmit over, the more infor-
mation you can send. This is why a low-fidelity AM radio station
transmitting over a bandwidth of only 5000 Hz (hertz) can’t
match the fidelity of an FM station using a bandwidth of around
100,000 Hz, not to mention a TV station operating on a band-
width of 6 MHz.

Since we’ll have to see the beacon before getting the message,
it seems likely that our receivers will need to have very fine reso-
lution, even down to 1 Hz. To see why, just imagine a beacon
broadcasting exactly at the waterhole frequency of 1420 MHz,
and suppose we had a receiver whose resolution was such that we
could distinguish signals separated in frequency by no less than
100 MHz. In other words, we could hear signals at 1300 MHz,
1400 MHz, 1500 MHz, and so on, but could not distinguish any-
thing in between. Such a receiver would pass right over the
magic frequency of the beacon, leading us down the garden path
of no ETIs, when in reality they’re just waiting with bated
breath to get in touch with us. Unfortunately, it takes a lot lon-
ger to search from 1400 MHz to 1500 MHz if you do it in steps
of 1 Hz than if you do it in one fell swoop, so most current
search strategies try to make some sort of compromise between
high resolution and search time.

SEARCH DIRECTION

In Project Ozma, Drake & Co. pointed their telescope at Tau
Ceti and Epsilon Eridani primarily on the basis of the G-star
chauvinism discussed earlier. Regrettably, the Principle of
Mediocrity comes into play here with the unhappy fact that the
universe is just filled with G-type stars. In fact, in almost any
direction you look there’s an abundance of stars of “our type,”

372

PARADIGMS LOST

so this requirement doesn’t really restrict the search space to
any appreciable degree. About all that can be said in this regard
is that all things being equal (as they never are), it’s probably a
good idea to stay away from the galactic center as there are all
sorts of events going on there, none of them conducive to living
the good life — or any life at all.

One interesting variant on the direction theme has been pro-
posed by Michael Papagiannis, a seemingly tireless astronomer
at Boston University who is also president of the International
Astronomical Union’s Special Commission 51 on Bioastronomy
(as SETI is euphemistically termed in polite scientific circles).
Papagiannis suggested that if ETIs existed and were in a colo-
nizing mood, the most likely place for them to take up abode in
our solar system would be in the asteroid belt between Mars and
Jupiter, since there they would find plenty of the raw materials
needed to sustain an exploratory colony. Consequently, his idea
is to focus some attention on looking for signs of ETI in our
own solar system, as well as searching the stars. Seeking ETI in
the asteroid belt may indeed be a stroke of divine inspiration.
But at the moment it appears that most telescopes are not
focused in this direction.

Before considering some of the actual searches that have been
conducted so far, it’s worth pausing to emphasize that we have
been talking here only about searches in the radio-frequency
part of the electromagnetic spectrum (10,000 Hz to 1000 MHz).
Some have advocated searches at other wavelengths, primarily
in the infrared 100,000 to 400 million MHz. The initial sugges-
tion along these lines came in a short note to Science in 1960 by
Freeman Dyson, who noted that a truly advanced civilization
would surely have developed the technology needed to harness
the entire energy output of its parent star. His suggestion was
that such a civilization would dismantle all the planets of its
solar system, using the matter to form a shell surrounding the
central star in order to prevent the escape of enormous amounts
of solar energy into outer space. Such a sphere would capture all
the solar radiation for use by the ETI civilization, and a by-
product of such capture would be that the sphere would radiate
strongly in the infrared part of the spectrum.

A civilization capable of the kind of engineering magic needed
to construct such a Dyson sphere is termed a Type II civilization
in the classification scheme of Nikolai Kardeshev, a Russian
SETI expert. According to this scheme, a Type I civilization is

WHERE ARE THEY?

373

one at a level of development similar to our own, capable of utili-
zing most of the energy of its own planet, while a Type III can
command the energy of an entire galaxy. By Dyson’s argu-
ments, we should tune our telescopes to the infrared part of the
spectrum to see signs of a Type II civilization. Of course, eaves-
dropping on the radiated energy of ETI and listening for de-
liberate signals are quite different matters, requiring corre-
spondingly distinct observing strategies. So at present there’s
not too much attention being devoted to looking for Dyson
spheres. Other proposals are even more fanciful, involving sig-
nals sent by beams of neutrinos, tachyons (faster-than-light par-
ticles), and other mechanisms that at present are more properly
left to the speculations of science fiction than to the realm of
science fact. Now let’s take a look at some of the searches since
Ozma in an attempt to understand how we might recognize an
ETI signal if we saw one.

WHAT ARE WE LISTENING FOR? — THE
SYNTAX AND SEMANTICS OF SETI

Suppose you’re a radio astronomer interested in SETI, and you
manage to talk your bosses into giving you a little bit of the
telescope’s “dead time” to indulge your curiosity. You decide to
do a “conventional wisdom” search at the waterhole frequency
of 1420 MHz, and turn your telescope to one of the likely G-type
stars in our galaxy, such as Tau Ceti. What exactly would you
see, and how could you tell if your record really did include a
signal from an advanced Tau Cetacean civilization?

To address this question, at the historic 1971 Byurakan meet-
ing I. S. Shklovskii presented the diagram shown in Figure 6.4,
which represents an artificially generated record of the type ob-
tained by a radiotelescope. The graph is formed by superimpos-
ing random noise upon eighteen weak signals. The places where
the signals occur are indicated by the small windows marked
along the top of the diagram. This example should easily con-
vince you that it’s not possible to detect the existence of an en-
semble of signals just by eyeballing the usual radiotelescope
record. But by making use of statistical techniques of cross-cor-
relation of the two independent spectra of the signal and the
noise, the pattern of Figure 6.5 is obtained. The rather marked
peak at reference delay 0 indicates the presence of a nonrandom
component in the original record, i.e., a signal. This is one of the

374

PARADIGMS LOST

standard procedures for recognizing that a real signal is embed-
ded within an otherwise noisy record.

Another way of checking for the presence of a signal, espe-
cially if it’s of the beacon variety, is just to point the telescope
directly at the star and measure the received energy, then point
it slightly away from the star and see if the energy from the
comparison point differs in any significant manner. Figure 6.6
displays an experiment of this type carried out on Tau Ceti by
Gerritt Yerschuur at the waterhole wavelength of 21 cm (i.e.,
frequency 1420 MHz). In this case, it’s clear just by inspection
that there is no real difference in the received patterns from the

WHERE ARE THEY?

375

On Star Comparison Point

FIGURE 6.6. The telescope record from Tau Ceti
star and from the comparison point, which is about 20 minutes
of arc away.

One of the most intriguing signals ever recorded is displayed
in Figure 6.7. This is the famous “WOW” signal recorded in
1977 by the Ohio State University SETI project headed by John
Kraus and Robert Dixon. The strength of the signals received in
each of the fifty channels of observation are recorded on the left
side of the figure, while the right side just indicates where in the
sky the telescope was pointing. Note that mostly the received en-
ergy can be represented by a small number, usually 1 or 2. How-
ever, the WOW signal was so strong that it was necessary to go
beyond the integers and use letters through Q to represent its
magnitude. Regrettably, the signal was never seen again, despite
repeated efforts to reacquire it by many investigators around the
world over the last decade. So for now it’s necessary to relegate
the WOW signal to the ever-growing category of heart attack-
inducing SETI anomalies.

Over the past several years, Jill Tarter of the NASA Ames
Research Center and the University of California, Berkeley, has
become the unofficial keeper of the books for ETI radio searches,
having at last count (in 1987) recorded forty-five such efforts
beginning with Project Ozma. So far there have been no suc-
cesses, although all theoretical arguments conclude that we
should expect to search for centuries before having a betting
man’s odds of actually finding an ETI signal, even if it is out
there. (In this regard, Frank Drake sets a search horizon of five
thousand years as an off-the-cuff estimate.) Nevertheless activity
has never been more feverish in this area, with NASA currently
in the process of putting together a decade-long SETI radio ef-
fort that should start around 1990. The NASA SETI program is
divided into two parts: an all-sky survey, which will look at the

376

PARADIGMS LOST

entire sky over a wide frequency range but with rather low sen-
sitivity, and a targeted search, which will take a narrow-band-
width look at about eight hundred stars over a restricted range
of frequencies bordering the waterhole. Figure 6.8 displays the
section of the cosmic haystack that the NASA SETI program
will look at.

Oddly enough, the Russians, who were very positive about
radio searches for ETI in 1971 at Byurakan, seem to have dis-
continued all efforts along these lines. Rumor has it that Shklov-
skii, who was head of the Astrophysics Division of the Soviet
Academy of Sciences, apparently became disillusioned about the
prospects for either the existence or the detection of ETI (it’s
not clear which), with the result that virtually all Russian radio
search activity seems to have stopped. However, Shklovskii’s
death in 1985 may reopen the possibility of the Russians’ rejoin-
ing the hunt.

But the NASA SETI program is far from the only search
being contemplated over the next few years. Paul Horowitz of
the Harvard-Smithsonian Project has made use of the explosive
developments in microelectronics to develop an 8.4-million-chan-
nel narrowband spectrum analyzer, enabling his Sentinel Proj-
ect to complete the equivalent of a hundred thousand years of

WHERE ARE THEY?

377

Sensitivity, W/M2
Z

OZMA listening in one minute! Despite this incredible technolog-
ical advance, in five years of listening, Horowitz jokes, “we’ve
discovered the Sun twice.” We still need years of searching to
cover just the NASA parameter set, showing how truly immense
the galaxy really is, and what a microscopic sliver of the hay-
stack the searches so far have actually looked at.

For those cost-conscious SETI consumers, it’s noteworthy
that the cost of developing Horowitz’s piece of equipment was a
paltry $95,000, while the operating budget for the project itself
is an anemic $20,000 per year. Both sums, incidentally, have
been provided by the Planetary Society, a nonprofit SETI orga-
nization formed by Carl Sagan and partially sustained by movie
mogul Steven Spielberg, who apparently is willing to invest at
least a bit of his E.T. proceeds into the search for real ETs. At
present, the Sentinel Project, along with the Ohio State SETI
program which has been going since 1973, are the only totally
dedicated SETI radio searches, not devoting telescope time to
any other purpose. All the others involve either piggybacking
SETI on the search for other astronomical phenomena, or grab-

378

PARADIGMS LOST

bing time on telescopes in the odd moments when they’re not
engaged in other work.

Despite the almost negligible costs of SETI, especially as com-
pared with multibillion-dollar particle colliders or a trillion-dol-
lar SDI system, SETI enthusiasts have faced a continual uphill
battle to obtain even the microscopic level of funds needed to
carry on the search. A poignant example of this problem is pro-
vided by the Ohio State project, which has been run on a cost-
only basis for almost fifteen years now. None of the personnel
from directors Dixon and Kraus on down have taken a cent of
salary for their time, yet they have somehow managed to keep
the search alive by dint of Yankee ingenuity and sheer hard
work. But even this noble effort almost came unglued a few
years back when a different university that owned the land on
which their telescope sits wanted to sell it to a property devel-
oper for transformation into a golf course! Fortunately Kraus,
Dixon, & Co. were able to avoid this close encounter of the golf
course kind, but only after a sustained media campaign rein-
forced strong appeals by the scientific community.

A somewhat similar situation occurred in 1981 when NASA’s
SETI program was axed from the budget by an extraordinary
legislative amendment proposed by that eternally vigilant
guardian of the public purse, Senator William Proxmire. Un-
fortunately Congress passed the amendment, thereby mobilizing
the U.S. SETI community in a lobbying effort to get the money
reinstated. Enter the Planetary Society’s biggest gun, Carl
Sagan, who had had earlier interactions with Proxmire and
thought of him as being a reasonable man despite his public
image to the contrary. So off to Washington went Sagan, who
listened to Proxmire’s argument, which, in essence, came down
to the standard AT = 1 argument that if ETI existed, we would
have seen it by now. Sagan’s rejoinder was to point out the enor-
mous importance of the factor L, the lifetime of technical civili-
zations, in the Drake equation, and the crucial importance of
SETI for discovering whether or not there have been other civil-
izations that have avoided self-destruction. It turned out that
Proxmire had never even heard this line of argument before,
and after reflecting upon the whole issue decided to drop his ob-
jection. The final battle came when the issue had to be taken up
on the floor of Congress. But by good fortune the film E.T. had
just been released at that time, and was grossing more box-office
receipts per day than the entire amount NASA was seeking to

WHERE ARE THEY?

379

look for E.T.’s real-life siblings. Given the glacial pace of con-
gressional deliberations, nothing happened for weeks and
months. But finally, on September 30, 1981, the last day before
the new fiscal year, Congress passed interim funding to keep the
country solvent and in the process also passed the budget for the
independent agencies — including NASA. Thus was American
SETI saved from legislative extinction.

Most of the foregoing SETI stories deal with the syntactic
aspects of radio searches, i.e., how we might recognize a signal if
we saw one. But what about the semantic component? Imagine
we actually had a live ETI transmission in hand. What would it
be likely to say? What sorts of messages might be contained in a
collection of pulses of the type that most searchers think will
compose an information signal? No one really has any idea, of
course, about what an ETI might think is important enough to
try to transmit across the galaxy. So most studies of the seman-
tics of SETI naturally tend to focus upon the kind of message
that we might want to send to them (another good example of
the irresistible anthropomorphic bias of most SETI work).

On the northern coast of Puerto Rico, near the town of
Arecibo, there is a natural dish-shaped hole in the rock over
1,000 feet wide. Inside this bowl sits the world’s largest radio-
telescope. So large, in fact, is this dish that its collecting area
exceeds that of all the optical and microwave telescopes ever
built in the history of man. Put another way, it would take
around 4 billion bottles of beer to fill the Arecibo bowl. In 1974
modifications were made to this telescope enabling it to transmit
a radio beam of unprecedented power, up to 20 terawatts
(1 terawatt = 1 trillion watts) for a short interval. As an inau-
gural test of these changes, it was decided to use the Arecibo
dish to transmit a signal to the edge of our galaxy informing
potential listeners that “we are here!” This pathbreaking signal,
composed of a sequence of 1,679 binary l’s and 0’s, was trans-
mitted in a little under three minutes on November 16, 1974, at a
frequency of 2380 MHz with a bandwidth of 10 Hz. Note that
this is not the waterhole frequency but is still in the low part of
the cosmic thermal noise curve of Figure 6.3. The actual se-
quence transmitted is shown in Table 6.2. What kind of infor-
mational content about we earthlings could be contained in such
a sequence of pulses?

The logic underlying the message is to assume that any receiv-

380

PARADIGMS LOST

0 0 0 0 00 l 0 1010 1000000000000 I O 1 0 0000 I 0 I 0
000000100 100010001000 100101 1001010101
0 10 10 101 0100100 1000000000000000000000
0000000000000000 1 1 OOCOOOOOOOOOOOOOOOO

1 10 10000000000000000000 1 I 0 1 0000000000
00000000 1 0 1 0 l 000000000000000000 1 1 1 1 10
OOOOOOOOCOOOOOOOOOOOOOOOOOOOOOO 1 10000

1 1 10001 I 00001 1000100000000000001 10010

OOOIIOIOOOI'OOOIIOOOOI '0>0l 11 110111 ; I
01 1 1 1 10 1 I 1 I i 0000000000000000000000000
0 100000000000000000 1 00000000000000000
00000000000 1000000000000000001 1 i 1 l 1 00
000000000001 1 1 1 10000000000000000000C0
001 10000’ 10000! 1 10001 ' 000 i 0000000 i 000
OOOOOO’OOOOi 10100001 10001 l 1001 >0101 l l

i 1 0 i i l i ' 0 i i 1 i 101 1 l 1 100000000000000000
000000000 i 000000 l '000000000 1000000000
00' '000000000000000 100000' 10000000000
1 I 1 1 1 1000001 <0000001 1 ’ l 10000000000’ 10
000000000000 1 00000000 l 00000000 i 00000 1
OOOO’OO 11 0000000 '0000000' 100001 1000000
10000000000 1 1000 10000 I 100000000000000
01 100! 10000000000000 1 1000 ! 0000 1 10000 0
00001 ’00001 ’000000100000001000000 1000
00000’ 00000’ OOOOCOO 1 1 OOCOGOGO'OGG ■ 000

00000 1 1 00000000 1 000 l 000000000 10000000

1 00000 ’ 0000000 1 0000000 1 0000000 1 000000
0000001 ’000000000' ’00000000 1 lOOOCCCOO

0 10001 1 1010 1 ’ 00000000000 i 0000000 ' 0000

0000000000 I OCOCO l 1 1 i 10000000000001 000
0101 1 1 0 1 00 i 0 l 101 I 000000 i CO i i 1 0 0 i 0 0 i i i

1 i l 10! 1 100001 1 1000001 1 0 i l 100000000010
100000’ 1 10' '00100000010:000001 i l i i i 00
1000000101000001 ’ 000000 I 00000 1 l 0 1 1000
00000000000000000000000000000000 1 1 I 00

000 ’ 00000000000000 1 1 l 0 to 1 OOOlO i 0 10 <0 I
O’OOl 1 ’000000000 10 l 0 1 0 : OOCCCOOOOCCCGC
0010 100000000000000 ! 1 1 i : COOOCCCOCGOGG

0001 1 1 1 1 I 1 I i 000000000000 i 1 i 0 0 0 C 0 0 0 i 1 1
0000000001 '000000000001 1 0000000 I 10’ 00
0000000 10 1 100000 1 100 1 10000000' >00' '00
OOlOOOiOiOOOOO'O'OOO'OOOOlOOOlOO'OOOi
00 l 000 1 OOOOOOOC ’ 000 1 0 i OCO 1 OOOCOCOOOOO
0 1 0000 1 0000 1 000000000000 1 000000000 1 00
000000000000 1 00 '0 ’ 00000000000 1 ' 1 '00' i
11101001111000

TABLE 6.2. The 1974 Arecibo transmission

ing ETI will soon recognize that the number of pulses is the
product of the two prime factors 23 and 73; i.e., the unique ex-
pansion of 1,679 into prime factors is 1,679 = 23 X 73. (Recall:
An integer is prime if it is divisible only by itself and by 1.)
Since every integer can be written as the product of primes in
exactly one way, the fact that 1,679 has only two prime factors
suggests that the signal is actually a code for the construction of
a two-dimensional picture. By breaking up the message into sev-
enty-three rows of twenty-three characters each, arranging each
row one under the other, and letting 0 stand for a blank with 1
being a dark space, a clever ETI would arrive at the picture
shown on the left in Figure 6.9, with its interpretation given to
the right.

Starting at the top, the first part of the message is a counting
lesson that describes the number system that is to be used. The
numbers 1 through 10 are written across the top in binary nota-
tion. Notice that each number has a “number label” associated
with it, both to indicate that it is a number and to show from
which direction it is to be read. The numbers 8, 9, and 10 are
deliberately written on two separate lines to show how numbers

WHERE ARE THEY?

381

■ ■

FIGURE 6.9. The Arecibo message

too large to be specified on a single line will be written later. The
rest of the message deals with various physical, chemical, and
biological features of life on Earth. In fact, these parts of the
message are exactly what we would like to know about ETI in
order to fill in some of the blanks in the Drake equation. The
message concludes with a description of the telescope that sent
it, shown as centered on the third planet with the number across
the bottom indicating that the telescope is 2,430 wavelength

382

PARADIGMS LOST

units (1,000 feet) wide, where the natural wavelength for ETI to
assume is that on which the signal was sent. All very simple,
logical, and straightforward — once you know the key. This sig-
nal was beamed into the heart of the globular cluster Messier 13,
a collection of 300,000 stars in the constellation Hercules, which
is about 25,000 light-years from Earth.

Those readers who dabble in amateur cryptology might be in-
terested in trying their hand at decoding the message string de-
picted in Figure 6.10. This is a message created by Frank Drake
to show what we might receive from a hypothetical ETI. The
principles are the same as for the Arecibo message, but don’t be
discouraged if you can’t decode it — to extract the message would
probably require a whole team of specialists. (Hint: The message
consists of a total of 551 binary pulses.) The solution can be
found in “To Dig Deeper.” But just in case ETI isn’t listening
in M13, or is perhaps on a trip visiting elsewhere in the Galactic
Federation, there have been other efforts to send a souvenir of
Earth to the stars.

Sometime in 1989, the Pioneer 10 space probe will pass the
orbit of Pluto and become the first human artifact to leave the
solar system, moving on a heading roughly toward the star Al-
debaran in the constellation Taurus. Shortly before the launch
on March 2, 1972, Carl Sagan and Frank Drake suggested that a
small plaque be attached to the probe as a symbolic message to
any wandering ETI who might bump into it on its way to Tau-
rus. Surprisingly NASA agreed to the proposal, and a six- by
nine-inch gold anodized aluminum plate with the display de-
picted in Figure 6.11 was attached to the probe. As one might
have expected from the use of figures representing a naked
human male and female, as soon as the design was made public
the lunatic fringe started a mail-in campaign accusing NASA of
trafficking in space pornography. One can only wonder whether
E.T., Jabba the Hutt, or the Blob would find the figures one bit
erotic! In any case, the chances are nil that Pioneer 10 will ever
enter another solar system, so the whole exercise was far more
symbolic than real anyway.

Heartened by the generally positive response to the Pioneer 10
plaque, and never one to miss a chance to promote SETI in the
public arena, Carl Sagan saw the launch of the Voyager 1 and 2
probes in 1977 as another opportunity to spread human cheer
and goodwill outside the solar system. Since there was much

WHERE ARE THEY?

383

11110000101001000011001000000010000010100
10000011001011001111000001100001101000000
00100000100001000010001010100001000000000
00000000001000100000000001011000000000000
000000010001 1101 10101 10101000000000000000
00001001000011101010101000000000101010101
00000000011101010101110101100000001000000
00000000000100000000000001000100111111000
00111010000010110000011100000001000000000
10000000010000000111110000001011000101110
10000000110010111110101111100010011111001
00000000000111110000001011000111111100000
10000011000001100001000011000000011000101
001000 111100101111

FIGURE 6.10.4 message from a hypothetical alien

Hyperfine transition of Silhouette of Binary equivalent

neutral hydrogen spacecraft of decimal 8

center of the galaxy

FIGURE 6.11. The plaque on Pioneer 10

384

PARADIGMS LOST

more time to prepare the message than for the Pioneer effort, the
Voyager communiques could be far more complex and imagina-
tive than just a simple plaque. As a result, both of the Voyager
probes carried a special kind of videodisk upon which was en-
coded much of our scientific information, as well as a medley of
earthly sights and sounds, sort of an interstellar version of “The
Earth’s Greatest Hits.” Table 6.3 gives a list of the contents.

It’s not without interest to note that Sagan seems to have had
a change of heart sometime after these exercises, since in a 1983
article in Science he argues rather strongly in favor of the cur-
rent SETI programs of listening instead of sending, basing his
case on the following points:

• New kid on the block: Since we’re just entering the SETI game,
few ET civilizations could be more backward than we are.
Hence, we should be listening, not sending.

• Poor mouthing: Civilizations considerably more advanced than
we are would have vastly greater energy resources and more
sophisticated technologies that they could use for transmis-
sion.

• Barbarism: Two-way conversations that may take centuries
have not yet entered into our long-term planning processes,
which mostly don’t extend beyond the next election or war.

• Hide-and-seek: By sending, we might “give our position away”
to an unscrupulous ETI who might want to plunder our
planetary resources or use us for slaves or food.

• Village idiot: It’s not clear that we have anything interesting
to say.

Most of these points are debatable at best, with the exception of
“hide-and-seek,” which is not even debatable since we gave away
our position years ago with the escape of our TY transmissions
of I Love Lucy, Dallas, and Mork & Mindy, as well as military
radar signal leakage outside the ionosphere. Nevertheless, the
fact remains that at present no one is worrying about sending,
all the programs being devoted to various forms of listening. My
own guess is that it’s a matter of pure economics. It’s hard
enough to talk the Proxmires of the world out of the few million
dollars that NASA spends on SETI each year. Imagine what
they’d say if you told them you were going to spend the money
and merely send information, not try to receive it. I rest my
case.

385

WHERE ARE THEY?

PICTURES (m sequence)

calibration circle

letus

dolphins

cotton harvest

factory interior

solar location map

diag ot male and lemale

school of tish

grape picker

museum

mathematical definitions

birth

tree toad

supermarket

X-ray ol hand

physical unit del mil ions

nursing mother

crocodile

diver with tish

woman with microscope

solar sys parameters (2)

lather and daughter (Malaysia)

eagle

fishing boat, nets

Pakistan street scene

the sun

group of children

S African waterhole

cooking tish

India rush-hour traffic

solar spectrum

diagram of family ages

Jane Goodall. chimps

Chinese dinner

modern highway (Ithaca)

Mercury

tamity portrait

sketch of bushman

licking, eating, drinking

Golden Gate Bridge

Mars

continental drift diagram

bushmen hunters

Great Wall of China

tram

Juprter

structure of earth

Guatemalan man

Alncan house construction

airplane m (light

Earth

Heron Island (Australia)

Balinese dancer

Anush construction scene

airport (Toronto)

Egypt. Red Sea. Sinai

seashore

Andean guts

Atncan house

Antarctic expedition

Pen . Nile (horn ortxl)

Snake River. Grand letons

Thai craftsman

New England house

radio telescope

chemical definitions

sand dunes

elephant

modern house (Cloudcrotl)

(Wester Dork)

DNA structure

Monument \Olley

Turkish man with beard

house interior with

radio telescope (Areobo)

DNA structure magnified

leal

and glasses

artist and lire

book page (Newton's Sys/i

cells and ceil division

(alien leaves

old man with dog and

lai Mahal

ol the World)

anatomy (8)

sequoia

flowers

English city (Oxford)

astronaut m space

human sex organs (drawing)

snowflake

mountain climber

Boston

Titan Centaur launch

conception digram

tree wifi daffodils

Cathy Rigby

UN building (day)

sunset with turds

conception photo

Hying insect, flowers

Olympic sprinters

UN building (night)

string quartet

fertilized ovum

vertebrate evolution drag

schoolroom

Sydney Opera House

viotm with score

letus diagram

seashefl (Xancidae)

children with globe

artisan with drill

GREETINGS IN MANY TONGUES (alphabetically)

Akkadian

English

Ha (Zamba)

Mandarin

Punjabi

Swedish

Amoy (Mm dal )

French

Indonesian

Marathi (inda)

Rajasthani

tougu (inda)

Arabic

German

Italian

Nepali

Romanian

Thai

Aramaic

Greek

Japanese

Nguru (SE Africa)

Russian

Turkish

Armenon

Gujarati (Indu)

Kannada (Inda)

Nyania (Malawi)

Serbian

Ukranon

Bengali

Hebrew

Kechua (Peru)

Onya (Inda)

Sinhalese (Sr. Lanka)

Urdu

Burmese

Hindi

Korean

Persian

Sot ho (Lesotho)

Vietnamese

Cantonese

Hittue

Latin

Polish

Spanish

Welsh

Czech

Hungarian

L uganda (Uganda)

Portuguese

Sumer on

Wu (Shanghai dal )

Dutch

SOUNOS Of EARTH (in sequence)

•tales surf

planets (audio cricket frogs

anaiofl o< turds

ortkttl velocity) hyena

volcanoes elephant

mul pots chimpanzee

ram wild dog

laughter

lire

tools

dogs (domestic)
herding sheep

blacksmith shop

sawing

tractor

riveter

Morse code

ships

horse and cart

horse and carnage kiss

Iran whistle baby

tractor tile signs

truck EEG. EKG

auto gears pulsar

Saturn 5 rocket
liltotf

MUSIC (m sequence)

Bach Brandenburg Concerto #2. 1st m
Java court ganelan— 'Kinds of Flowers*
Senegal percussion
Zaire: *F*ygmy girls' initiation song
Australia horn and totem song
Mexico manachi— *EI Cascabei*

Chuck Berry 'Johnny B Goode*

New Gmnaa men s house
Japan shakuhacfc (flute)—

'Depicting the Cranes m Their Nest*

Bach Partita #3 tor violin
Mozart "Queen ol the Night*

(from 'The Magic Flute*)

Georgia (USSR) totk chorus— 'Chakrulo*
Peru pan pipes

Louis Armstrong 'Melancholy Blues'
Azerbaijan two (lutes
Stravinsky 'Rile ol Spring* conclusion
Bach Prelude and Fugue #1 m C Maior
Beethoven Symphony #5. 1st m

Bulgaria shepherdess song—

"iztel Detyo hajdutin*

Navaio night chant

English 15th cent 'The Fame Round*

Melanesia: pan pipes

Peru woman's wedding song

China ch in (zither)— 'Flowing Streams*

India raga— *Jaat Kahan Ho*

Blind Willie Johnson 'Dark Was the Night*
Beethoven String Quartet #13. 'Cavatina*

TABLE 6.3. The contents of the Voyager disk

386

PARADIGMS LOST

* « *

Having now considered the main theoretical and experimental
underpinnings of the ETI question, it’s time to let the ideolo-
gists of the N = 1 and the N > 1 schools have their day in
court. But before entering the courtroom and listening to their
respective arguments, let’s try to summarize the various subdi-
visions of the problem by listing ten possible answers given by
astronomer John Ball to the original Fermi question: “Why are
we unaware of ETI?”

1. There is no ETI. Either Earth is unique or ours is the first
civilization in the galaxy to reach this level of development.

2. ETI exists, but it’s very primitive. It doesn’t know we’re here,
but it might like to know.

3. ETI exists and is at about our level of development. It suspects
we might be here and it might like to talk with us (the Mir-
ror Hypothesis).

4. ETI exists and it knows we’re here. It would like to talk with
us if it could just attract our attention.

5. ETI exists and knows we’re here, but it doesn’t care. We pose
no threat and we have nothing that it wants.

6. ETI exists and we are of some interest to it. A few ETI scien-
tists are discreetly studying us.

7. ETI exists and we are of considerable interest to it. It is study-
ing us in some detail, but inconspicuously.

8. ETI exists and it occasionally dabbles in our affairs. We are of
considerable interest to ETI and it wants to interact with us
directly (the UFO Hypothesis).

9. ETI exists and is experimenting with us. We are laboratory
animals for it (the Petri Dish Hypothesis).

10. God exists. A supernatural being who is omnipotent and om-
niscient exists (i.e., God is identical with ETI).

All of these views except the first imply the existence of ETI,
although Cases 2 through 9 are not mutually exclusive. Cases 1
to 4 are the popular views, with 2, 3, and 4 representing the
dominant view of the SETI scientific community. Cases 6 and 7
are commonly termed the Zoo Hypothesis, for obvious reasons.
Beginning with Case 8, one leaves the realm of science and en-
ters into the domain of religion and philosophy, Case 10 being
the popular nonscientific position.

Since what we’re interested in here is science, let’s lump Cases

WHERE ARE THEY?

387

3 through 7 under the general label N > 1, while the other side
of the trial, N = 1, will be associated with Case 1. Case 2 in-
volves ETIs so primitive or profoundly alien that no communi-
cation is yet possible, so I also lump this case in with the N = 1
side of the house. Now, having completed the preliminaries, let’s
listen to the Prosecution arguments for why we are not alone in
the galaxy.

N > 1: ETI EXISTS!

The early 1970s were a particularly cordial period in U.S.-So-
viet relations, when even the chronically overbooked Moscow
“gourmet restaurant” Aragvi was always ready to accommodate
a “famous visiting American professor” by mysteriously con-
juring up a table without benefit of the traditional na leva gra-
tuity, an opportunity I myself was always ready to exploit
during a 1972 tour of duty at the Control Sciences Institute of
the USSR Academy of Sciences. From September 5 to 11, 1971,
during this all-too-brief golden age of detente, the Byurakan
Observatory near the Armenian capital of Yerevan hosted what
is still one of the most extraordinary SETI meetings ever con-
vened. This Soviet- American gathering at the foot of Mount
Ararat, which we mentioned briefly earlier, had as its unofficial
agenda a detailed analysis of each of the terms in the Drake
equation, together with a consideration of the various experi-
mental attacks upon ETI as outlined above. While the experi-
mental state of ETI research has improved by several orders of
magnitude since this historic event, a reading of the transcript
shows that the theoretical speculations are still as fresh and
timely as the day they were proposed over fifteen years ago (a
good indicator of the level of hard data versus soft speculation
in the theoretical ETI business).

After a week of “Armenian breakfasts,” an indispensable in-
gredient of which is a shot or two of the fiery local cognac, the
theoretical underpinnings of the entire N > 1 school of SETI
thought were laid down, mostly by an American contingent nick-
named the Cornell group by the distinguished historian William
McNeill, himself a meeting participant. This constellation of
SETI devotees consisted of Carl Sagan, Philip Morrison, Frank
Drake, and Thomas Gold, all of whom were or had recently been

388

PARADIGMS LOST

on the faculty of Cornell University at the time of the meeting.
The essence of the position put forth at Byurakan was that by
inserting their best scientific estimates, subjective probabilities,
and just plain hunches into the Drake equation and turning the
crank, the outcome would be a number N far greater than 1.
Since this line of argument has already been addressed in some
detail above, let me try to summarize the core of the N > 1
thesis using the following chain of reasoning:

Every shred of genuine scientific fact points to the conclusion
that the Earth and our solar system are perfectly ordinary and
typical in every possible way (the Principle of Mediocrity).

Since life, intelligence, technology, and all the rest have
developed here on Earth, in the absence of further information
we must assume that these conditions are typical
elsewhere as well.

THEREFORE

ETI exists elsewhere in our galaxy; i.e., N > 1.

As a corollary to the claim that N is greater than 1, it’s of
interest to remark upon the final resolution of the meeting,
which, incidentally, serves as an exemplary model for East-West
scientific cooperation and goodwill. It states in part that “the
Conference participants . . . agreed that the promise of contacts
with such extraterrestrial civilizations is sufficiently high to jus-
tify initiating a variety of well-formulated search programs.”
Thus was the manifesto of the N > 1 enthusiasts laid down, and
thus it stands to this day: The likelihood of N being larger than
1 is sufficiently great to justify the costs of actively searching.
In following the literature over the years since Byurakan, it’s
intriguing to see just what kinds of proposals for “searching”
the Byurakan declaration has generated.

While the majority of scientists concerned with SETI have
understandably concerned themselves with the sorts of radio
searches considered earlier, there have been the usual extremists
at both ends of the scientific spectrum who interpreted the word
“search” in the literal sense, and focused their energy, calcula-
tors, and typewriters on the matter of direct contact. These vi-

WHERE ARE THEY?

389

sionaries fall into two totally distinct groups: the UFOers and
the space travelers. Since there has never yet been an unambigu-
ously documented case of an extraterrestrial visit to Earth, I
won’t open this can of worms here, leaving those who feel a deep
psychological need to believe in direct extraterrestrial interven-
tion in our puny affairs free to do so. Instead, in what follows
I’ll settle for considering some of the less contentious scientific
arguments for space travel as a means of contact.

First of all, the various Apollo, Viking, Voyager, and Pioneer
programs, as well as similar Soviet ventures to Venus and more
recently to Mars, leave little doubt that space travel of at least a
limited sort is well within the realm of our current technology
and purse. The problem for SETI is that by now it’s generally
conceded that there are no intelligent life forms on any of the
planets of our solar system, implying that if we want to meet
ETI face to face, we’re going to have to wind up our big toys
and set off into the interstellar void. Just how technically and
economically feasible is it to mount an expedition to visit even
one of the “nearby” stars?

To get some feel for the magnitude of the problem, think
about the distance involved in traveling to the Moon. The Moon
is about 240,000 miles from Earth and represents the greatest
distance that man has yet ventured beyond Earth. If we imagine
this distance as being the same as walking across your living
room, then on the same scale a trip to the nearest star is equiva-
lent to going to the Moon. In other words, such a trip represents
about 100 million trips to the Moon! And this is only to the near-
est star, Alpha Centauri, which, unfortunately, is not very inter-
esting from an ETI standpoint. To get to Ozma’s candidates,
Tau Ceti and Epsilon Eridani, each around 11 light-years away,
would involve about 300 million such trips. These are not dis-
tances to be taken lightly (no pun intended). So distance alone
imposes severe restrictions on what can be done about going out
to find ETI.

But let’s suppose, as some have done, that it’s feasible to con-
struct a ship using some sort of superduper fusion or antimatter
drive that will allow a ship to travel at 0.1 c , one-tenth the speed
of light. Studies have shown that no new physical principles are
involved in making such a vision a reality, although the engi-
neering hurdles are enormous. At such a velocity, you might be
able to visit one of the stars shown in Figure 6.12 within your

390

PARADIGMS LOST

lifetime. Techniques of suspended animation could extend this
limit, but probably not by much without introduction of major
new, and totally unpredictable, biological and physical princi-
ples. So on physical grounds alone the prospects for a generation
of Captain Kirks venturing where no man has gone before ap-
pear rather bleak, at least if that venturing extends much
beyond a few light-years from Earth. But let’s imagine that all
the physics and engineering works out and you’re determined to
check up on doings at Wolf 359 or Procyon. How much would it
cost to indulge your curiosity?

The question of economics ultimately comes down to how
much energy you need in order to transport yourself and your
belongings to a nearby star. Some back-of-the-envelope calcula-
tions are sobering. Let’s suppose that E represents the amount
of energy needed to maintain what you might think of as the
“good life.” Assuming as above that you can travel at 0.1 c and
that you require 10 tons of mass in your spacecraft per passen-
ger (about the same ratio as for a large jet passenger plane), the
amount of energy needed for our prototypical 100-year trip is
about 2 million X E. To pin down a value for E, let’s consider
the annual energy consumption in the United States. In 1979 this
figure was 1020 joules, leading to a value for E of 4 X 1013 joules ( 1

WHERE ARE THEY?

391

joule = the amount of energy needed to raise the temperature of 1
cubic centimeter of water by about j°C). Putting these figures to-
gether, we come to the sad conclusion that the minimum energy
needed for one passenger to Tau Ceti is around 8 X 1019 joules. Thus,
for a 100-passenger colony this represents an amount of energy
sufficient to sustain the entire American population, the most profli-
gate in history, for a period of several hundred years. The bottom line
of this elementary calculation is that unless there is some develop-
ment that makes energy literally cost-free, no society will ever be able
to underwrite the cost of sending you on your journey to the stars.

But just to keep the fiction alive, let’s suppose such a free
energy source was miraculously discovered and we set sail in
search of ETI. What might we find? Would ETI be a cute little
wide-eyed charmer like E.T., or would it be more like the night-
marish creature of A lien, or perhaps neither? And what kind of
social order might ETI have developed to enable it to survive the
various perils of postindustrial life outlined earlier? These are
some of the issues that ETI theoreticians of the N > 1 persua-
sion have fun speculating about when their real day’s work is
done. It’s impossible not to be sucked in by this kind of specula-
tion, so let’s carry the A^ > 1 claim to ridiculous extremes and
consider a few of the more sober, or at least scientifically defen-
sible, possibilities for the form of ETI.

THE SHAPE OP ETIS TO COME

Most of the reasons put forth by the SETI community as jus-
tifications for trying to make contact with ETI are of a some-
what lofty and depressingly sober character, viz., renewed hope
in knowing that another civilization managed to survive our cur-
rent market basket of nuclear, ecological, and psychological
crises, membership in a galactic federation, technological won-
ders such as free energy, teleportation, and immortality. Such
noble aims are just the ticket for scientists, sages, and congress-
men, but as for myself these supposed benefits of SETI are only
moderately interesting, and my own interest in ETI is distinctly
more visceral — I want to know what it looks like! Direct contact
is the best way to scratch this itch, although radio searches
might supply the information if the message is one of a pictorial
nature like those considered earlier, or if the message gives in-

392

PARADIGMS LOST

structions about how to construct a living ETI here on Earth.
But in the absence of any contact, direct or otherwise, we have
to fall back upon our own biological knowledge to speculate on
what might step out of that first UFO to become an IFO. There
are two diametrically opposed views on this matter.

The first line of reasoning about the form of ETI is to argue
by appeal to the process of convergent evolution. On Earth, when-
ever Nature has had a problem to solve, whether it be the opti-
mal design for a sense organ to process visible light or an
efficient way to tear food apart into bite-sized pieces, there has
been a tendency for the problem to be solved in a very similar
manner across a wide variety of species. Movement in water is a
good example. In the course of Earth’s history, there have been
three species that made their living by swimming rapidly in
coastal waters preying on small fish to fill their stomachs. These
three species are the tuna (a fish), the dolphin (a mammal), and
the ichthyosaur (an extinct reptile). These three animals have
very little to do with each other, biochemically or phylogeneti-
cally. But if we examine their physical forms, we find they all
look about the same — like a living torpedo. This is a good exam-
ple of how several different evolutionary paths may “converge”
to the same ecological position. It just happens to be very effi-
cient to have a torpedo-shaped body if you need to swim fast to
catch your dinner. The convergent evolution school applies this
same general principle to speculating about ETI. We can sum
up the convergent evolution thesis in the following diagram:

Intelligence > Communication + Mobility

v v v

Body and nervous Sensory organs Limbs

system

l' V

Large brain Land dweller

Life on earth has evolved two distinct types of symmetry, bi-
lateral and radial, and it’s no accident that the most successful
life forms have bilateral symmetry. As we’ve noted, it appears
very likely that life evolved in the oceans, and in such a watery
medium an organism with a streamlined body has a distinct com-

WHERE ARE THEY?

393

petitive advantage when it conies to catching prey or escaping
from predators. On the other hand most life forms with radial
symmetry lead rather starfishlike sedentary lives and have sim-
ple nervous systems.

It further appears that a necessary precondition for the devel-
opment of a complex nervous system is an active, mobile, preda-
tory life-style. In such a predatory life form with a complex
nervous system, the central controlling brain should be close to
the primary sense organs so that the connecting nerve paths are
short and the animal’s response time is correspondingly fast.
Such an animal must also have its sensing and grasping organs
at the front of the body near the mouth and, if it must smell its
food before eating, the organ for this sense must be located
above its mouth. Thus bilateralism and the presence of large
ganglia of nerves near the front of the body and close to the
primary sense organs are essential characteristics of intelligent
creatures in the convergent evolution scheme of things.

On Earth, birds, fish, and mammals all conform to the above
requirements. However, birds are not likely to develop high in-
telligence because they must be light and have a large surface
area in order to fly. Thus they can’t afford the weight of a large
brain and the heart needed to supply such a brain with enough
blood to keep it going. Life in the water doesn’t suffer from these
drawbacks, as witness, for example, the whales, which are the
largest creatures ever to exist on this planet. However, it can be
argued that life in the water is too easy to afford the kinds of
challenges to survival necessary to stimulate development of a
complex nervous system. That is, challenges of the type leading
to higher brain functions are usually associated with three
things: the use of tools, the development of language, and the
formation into social groups. Only land-based mammals fulfill
all of these conditions.

As a final exemplar of convergent evolution, there is the devel-
opment of jointed legs, which seem to be the best solution for
moving over different types of terrain. But a large number of
legs make for difficulties in coordination and slowness of move-
ment, while an odd number would create an awkward imbalance.
Thus, the swift runner would probably have only a small num-
ber of legs in pairs, of which one or two pairs would possibly
have been modified to act as arms for the manipulation of tools.

Putting all these considerations together, one comes up with

394

PARADIGMS LOST

an ETI whose physical form would be remarkably humanoid; in
fact, remarkably like the kinds of forms reported by people who
are abducted by the occupants of UFOs, or maybe like the be-
nevolent aliens depicted in the film Close Encounters of the Third
Kind. In all cases these ETIs look just about like you and me,
usually with the exception of a more pronounced egg-shaped
skull, presumably an indication of the far more advanced state
of their cerebral development.

Frankly I find this sort of anthropomorphic argument totally
unimaginative and quite boring. What a delicious cosmic joke it
would be to spend a few zillion dollars to look in on happenings
around Barnard’s Star, and find a planet where everyone drove
Fords, ate at McDonald’s, and watched The Coshy Show! But for
alternatives it’s not sufficient to accept the standard arguments
that there may be many different pathways to creatures func-
tionally equivalent to, but physically unlike, ourselves. As one
might suspect in matters of the imagination, in a search for al-
ternatives we have to leave the mainline scientific community be-
hind and turn to the science fiction writers and philosophers for
some mind-bending, yet physically feasible, candidates.

One of the great depictions of an alien life form in fiction is
given in Donald Moffitt’s novel The Jupiter Theft, which tells of
the plight of the Cygnans, a race of creatures that evolved on the
satellite of a gas giant planet orbiting a binary star system. One
of the members of the stellar pair collapses into a black hole, but
the Cygnans have sufficient warning that a small colony escapes
in five 30-mile-long ships, the interiors of which are primarily
huge, open, artificial forests where the Cygnans live alongside
the small arboreal animals they catch for food. The story tells of
how these creatures enter our solar system and begin disman-
tling Jupiter as a material source of energy, and the feeble at-
tempts of humans to do something about it. Figure 6.13 is an
artist’s interpretation of how the Cygnans look. The Cygnan is
about 1.5 meters tall, with six limbs that can be used as either
arms or legs, and a long, three-petaled tail that folds to conceal
the sexual organs. The slender, tubular body is built on a car-
tilaginous skeleton, with the brain located between the upper
pair of limbs at the top of the spinal cord. The three eyes are
placed on stalks in an equilateral triangle around a broad, flexi-
ble mouth. The Cygnan has a harsh, rasping plate in the mouth,
and a spiked, tubular tongue. It has a well-integrated nervous

WHERE ARE THEY?

395

FIGURE 6.13.4 Cygnan from The Jupiter Theft

system, with much faster synaptic reflexes than those of a
human being. Cygnan speech is musical, consisting of chords
produced by multiple larynxes, and depends on absolute pitch.
The language is incredibly rich and varied; it has more than a
million phonemes, and each word is made up of several pho-
nemes. But if you think the Cygnans are still too humanoid,
let’s look at another possibility.

On Earth, the only intelligent entities that differ radically
from the kind of bilateral humanoid considered earlier are colo-
nies of social insects like bees and termites. Arguments have
been made that ETI life forms might also adopt this kind of
bottom-up mode of group intelligence. A sci-fi example is de-
picted in Figure 6.14.

This picture shows the Cryer, a creature from Joseph Green’s

396

PARADIGMS LOST

FIGURE 6.14. The Oryer from Conscience Interplanetary

book Conscience Interplanetary. The Cryer is an independently
functioning unit of a planet-wide silicon-based plant intelligence
inhabiting the planet Crystal, which has an atmosphere of 18
percent oxygen, the rest being nitrogen and hydrogen. Life on
Crystal is based on silicon, with a high percentage of metallic
elements. The Cryer resembles a two-meter-high bush with a
crystal-and-metal trunk and branches, with small, sharp glass
leaves. The trunk contains silicon memory units, powered by a
low-voltage solar storage battery and connected by fine silver
wires. About six feet up the Cryer’s trunk is an organic air-
vibration speaker membrane created for it by the planet-wide
entity to enable it to speak with human beings. It is a broad,
saucer-shaped leaf held in place by stretched wires to provide a

WHERE ARE THEY?

397

vibrating diaphragm. A magnetic field generated in silver wire
coils hanging on either side of the speaker causes it to vibrate to
produce sound.

The planet-wide intelligence consists of thousands of smaller
units like the Cryer, connected by an underground nervous sys-
tem of fine silver wire. Each unit has a specialized function,
some storing electricity generated by sunlight, some extracting
silver for constructing the nervous system, some providing mem-
ory storage, and some acting as sensor units. The overall intelli-
gence is able to perceive temperature, motion, position, electrical
potential, and vibrations through its member units.

Cygnans and Cryers give only a small taste of the kinds of
ETIs that may be out there, if one is to believe the science fiction
writers’ union. I bring them up here only to show that the argu-
ment from convergent evolution, while scientifically defensible
on the grounds that it’s happened at least once, is far from the
last word on the matter of the kind of physical form ETI might
assume. As to how ETI might act, we have already spent a chap-
ter examining the degree to which human actions are biologically
determined, coming to no definitive answers. So when it comes to
ETI’s actions, I think discretion is the better part of valor. Con-
sequently I’ll now return to the courtroom and listen to the
claims that there is no ETI and that the above science-fiction
possibilities are just that — fiction.

ETI? THERE’S NO SUCH THING: N = 1

Alfred Adler was one of the giants of modern psychoanalytic
thought, a onetime associate of Freud’s and the originator of the
notion that compensatory mechanisms are often developed to
combat what is now termed an inferiority complex. In 1974 an-
other Alfred Adler, a man who in my opinion could do with a
few sessions on the couch himself, published an absolutely hilari-
ous article in The Atlantic in which he takes the Byurakan meet-
ing as a vehicle for expounding an evidently deep-seated sense of
resentment against what he terms “the modern technologist.”
After soundly denouncing most of the Byurakan speculations
(which had continually been advertised as such by the speakers
themselves) as “lunatic assertions” and “intellectual pollu-
tants,” Adler goes on to note that “the human qualities most

398

PARADIGMS LOST

displayed by the conferees were . . . cupidity, inanity, and trivi-
ality.” At this point the article moves into high gear with its
main message: the nature of the technological mind as seen by
Professor Adler. According to this vision, “The modern tech-
nologist is a gifted, highly trained, opportunistic, humorless,
and unimaginative ass.” A couple of sentences later we learn
that “none of his fatuous pseudo-science is science; all of it is
empty of intellectual content, inflated with self-importance, and
held accountable for nothing.” Does this go for all modern tech-
nologists? My last employer definitely, but to condemn all mod-
ern technologists seems a bit much even to my cynical eye. And
whom does the ever cheerful Adler choose as the focal point of
his vivid invectives? None other than Johnny Carson’s SETI
consultant, Carl Sagan, who even in 1974 was already becoming
the lightning rod for discharge of the petty resentments and
jealousies of a host of less visible (and less talented) scientists,
disaffected writers, and academics.

This whole Adler business would hardly even be worthy of
mention if it were not for the fact that the article displays in a
particularly blatant manner the sophomoric attitude toward sci-
ence and scientists held in certain quarters of the academic com-
munity. But what’s more important for our purposes, it serves
as a symbolic opening salvo fired against the initial euphoria
emerging from the Byurakan sessions regarding the likelihood
of contacting ETI. While it’s hard to imagine anyone taking
Adler’s arguments as anything other than dark grumblings and
light entertainment, by the mid-1970s a backlash against the
F > 1 claims was definitely in the air, with the resulting fallout
threatening for a while to destroy even the small and tenuous
foothold that SETI had carved out for itself in the remote foot-
hills of the mountainous terrain of mainline science.

Arguments claiming N — 1 tend to come packaged in one of
two quite distinct wrappers: factorization and observation.

• Factorization: Here we find all arguments centering upon one
or more of the terms in the Drake equation. All that’s needed
to show that N is negligible is to demonstrate conclusively that
one term of the Drake equation is close enough to zero that,
for all practical purposes, it is zero. This is the goal of the
factorization artists — to produce a knockdown argument
showing just this, focusing upon one of the astrophysical, bio-

WHERE ARE THEY?

399

logical, psychological, or sociocultural terms in the equation.

• Observation: The observers use a quite different line of reason-
ing, the classical reductio ad absurdum, which goes as follows:
Suppose ETI does exist. What observable consequences would
be likely to follow from this assumption? Do we actually ob-
serve any of these consequences? If not, then it’s highly likely
that N = 1.

Let’s look at each class of claims in turn.

In 1975 the disaffection with the prevailing N > 1 attitude
toward ETI foreshadowed in Adler’s outburst was given scien-
tific form by Michael Hart, a young astronomer at the National
Center for Atmospheric Research in Boulder, Colorado. Hart,
now at Anne Arundel College in Maryland, and probably the
only practicing astronomer who also holds a degree in law, took
a Talmudic view of the ETI question, zeroing in on the one hard,
incontrovertible fact surrounding the whole issue: There are no
intelligent beings from outer space on Earth right now. His path-
breaking paper, titled “An Explanation for the Absence of Ex-
traterrestrials on Earth,” offers a detailed consideration of this
empirical observation, termed Fact A in the paper, concluding
that the most reasonable explanation for Fact A is that there are
no other advanced civilizations in our galaxy. It’s illuminating
to consider Hart’s analysis of Fact A in more detail.

In good, logical, legal fashion, Hart divides the possible expla-
nations for Fact A into five categories:

• Physical: Some kind of physical, biological, astronomical, or
engineering difficulty makes space travel unfeasible.

• Sociological: ETIs have not arrived because they have chosen
not to. This includes all explanations involving lack of inter-
est, motivation, or organization, as well as political obstacles.

• Temporal: ETIs have arisen so recently that they haven’t had
time to get here yet, even though they want to visit us.

• Historical: ETI has been here in the past, but is not here now.

• Uniqueness: There are no other civilizations in our galaxy. If
there were, Hart says, they would have colonized the solar sys-
tem a long time ago, and we would not be asking, “Where are
they?”

Hart dismisses the physical explanations by asserting that the
usual arguments against space travel involving time of travel

400

PARADIGMS LOST

and energy requirements are vastly overstated. It’s interesting
to note here that according to his calculations, the energy needed
to accelerate a ship to 0.1 c and decelerate it requires that the
ship carry about nine times its own weight in fuel. This calcula-
tion should be compared with the much later, and far more pes-
simistic, estimates of Drake considered earlier under what, in
my opinion, are far more realistic assumptions. Hart here also
dismisses other possible physical hazards, e.g., the danger of col-
lision with meteorites (traveling at 0.2 c, a 4-ounce rock will im-
pact a ship with the force of a 40-kiloton bomb, twice the force
of the atomic blast that leveled Hiroshima), cosmic rays, and so
forth.

As to sociological explanations for Fact A, Hart has the uni-
form argument that no sociological explanation will suffice un-
less it can be shown that the same argument will apply to every
race in the galaxy at all times. So if you think that ETI is not
here because it blew itself up in a nuclear Gotterddmmerung, then
you must show that every ETI that ever existed also blew itself
up. Hart claims that this argument is universal, and can be used
to counter any of the sociological explanations presented for
Fact A.

To address temporal explanations, it’s necessary to estimate
how long it would take ETI to reach us in a wave of coloniza-
tion. Hart calculates that with a ship velocity of 0.1 c, such an
expansion wave would move across the entire galaxy in about 2
million years. But the age of our galaxy is on the order of
10 billion years, so to accept the temporal explanation it’s neces-
sary to assume that it took 5,000 time units (1 time unit = 2
million years) for the first civilization to emerge that had the
inclination to explore the galaxy, but that the second such spe-
cies (i.e., mankind) arose less than 1 time unit later. Hart con-
cludes that while the temporal explanation is theoretically
feasible, it should be considered highly unlikely.

There are several versions of the historical explanation, the
most common being that ETI was here rather recently (less than
five thousand years ago) but didn’t hang around. The weakness
in this explanation is that it doesn’t explain why Earth was not
visited earlier. On the one hand, if ETI could have visited us
earlier, then we need a sociological explanation for why it didn’t;
on the other hand, if ETI visited us as soon as it was able and
this was only within the last five thousand years (only one four-

WHERE ARE THEY?

401

hundredth of a time unit), then this requires an even more re-
markable coincidence than that mentioned earlier in connection
with the temporal explanation. Another version of the historical
explanation is that the Earth was visited long ago, say over 50
million years in the past. The problem here is that one again
needs a sociological explanation to show why, in all the interven-
ing years, no other ETI ever came to Earth and stayed.

The sum total of Hart’s arguments comes down to a collection
of reasons why the final four alternatives are even less likely
than the uniqueness explanation for Pact A. Thus follows
Hart’s assertion that N = 1, ushering in a period of critical
scrutiny of the whole SETI enterprise from every point on the
scientific compass.

Within a year of the appearance of Hart’s claims, counterar-
guments were put forth by Laurence Cox, who appealed to the
principle that in order for any civilization to enter the coloniza-
tion game, it would be necessary for it to stabilize its population.
If not, then even at a population growth rate equaling our own,
the population of such an ETI society would quickly outstrip
all the colonizable planets in the galaxy. Under the hypothesis
that the society could solve the population explosion problem,
Cox calculates that the temporal explanation is the most likely
way to account for Fact A, and that the ETIs just haven’t yet
had time to reach us.

An intriguing variant of the Hart argument was put forth in
1980 by another of the Young Turks in the anti-ETI camp,
Frank Tipler, a mathematical physicist at Tulane with a pen-
chant for contentious views on modern cosmology. Tipler
launched a broadside against what he called the semireligious
overtones of the entire SETI program by arguing that any civi-
lization much more advanced than ours would surely be able to
construct the types of self -reproducing machines that were men-
tioned in our discussions of artificial life in Chapter Two. The
idea of sending probes to search the galaxy for signs of emerg-
ing technological life was first put forth in the 1960s by Stan-
ford astronomer Ronald Bracewell, who suggested that any
advanced civilization would surely choose this cost-effective way
of exploration in preference to direct travel. Such devices,
termed von Neumann probes, would represent an extremely cheap
way of exploring outer space, being able to cover the whole gal-

402

PARADIGMS LOST

axy for a few billion dollars. Star Trek: The Motion Picture was
based on the use of such machines, which are really nothing
more than very souped-up versions of the kinds of primitive
probes that we have already sent to the Moon, Mars, and Venus,
as well as to other bodies in our own solar system. In his article
“Extraterrestrial Intelligent Beings Do Not Exist,” Tipler
strongly argues the same point as Hart, that an expanding wave
of colonization by such probes would fill the galaxy in a time
much shorter than the current lifetime of the galaxy. Yet we
don’t see even the faintest sign of such a von Neumann probe;
hence, they don’t exist, and neither does ETI.

As a fascinating commentary on the sociological ways of mod-
em science, Tipler later published what he claims is the clearest
evidence for a “save-the-world, semireligious motivation” under-
lying mainstream, establishment SETI. The “evidence” he pre-
sents involves the treatment that pro-ETI reviewers gave his
critical paper. It seems that a shortened version of the full
paper was submitted to the prestigious journal Science , whose
editor sent it to Carl Sagan for review. Apparently Sagan re-
jected the paper for what Tipler saw as at least superficially
valid and relevant reasons. Tipler proceeded to revise the paper
to answer (to his satisfaction, at least) the objections raised by
Sagan, and then submitted the revised paper to the well-re-
spected astrophysical journal Icarus. As fate would have it, the
editors of Icarus also sent the paper to Sagan for refereeing,
with the result that it was again rejected with a referee’s report
identical to that earlier submitted to the editors of Science.
While it’s difficult to know the precise circumstances surround-
ing this particular case, the general phenomenon is well known
to any denizen of the academic deep, as few pagan rites quite
equal the atavistic, troglodytic satisfaction of setting your col-
league’s ego on its ass — the equivalent in academic circles of
what on the gridiron is known as The Sack. As Tipler tells it,
“Had Sagan rejected the paper with a claim that my changes
were inadequate, or asked someone else to referee the paper (and
reply to my changes), I would have, of course, disagreed with
the rejection, but I would have felt the rejection was based on
scientific grounds. As it is, I feel as if I have become involved in
a theological debate.” Tipler also recounts similar adverse re-
marks from Philip Morrison, who commented on how imprudent
it would be to abandon radio searches, a matter not even men-
tioned in Tipler’s article.

WHERE ARE THEY?

403

Of course, Sagan and Morrison represent the bastions of
SETI and the scientific establishment, and one has to keep in
mind that these attempts to keep Tipler out of print were taking
place in the period when SETI advocates were having their
funding problems with Congress. At that time, the last thing the
pro-ETI community wanted was for some young upstart from
Tulane to place a loaded gun in William Proxmire’s hands by
publishing a difficult-to-rebut argument in a well-respected and
widely circulated American scientific journal. Consequently, Ti-
pler’s article finally appeared in a British publication, The Quar-
terly Journal of the Royal Astronomical Society, an eminently
reputable journal but hardly coffee-table reading matter for con-
gressmen or their aides.

The moral of this strange little tale is only that scientists are
no more selfless than anyone else when it comes to recognizing
what side their bread is buttered on. And when matters come
down to the eternal struggle between the ideology of science and
the realities of economics, as Damon Runyon once remarked,
“The race is not always to the swift, nor the battle to the
strong — but that’s the way to bet.” In modem science, as in the
rest of modem life, ideals and ideologies are pretty feeble com-
petitors when they come into conflict with the pocketbook.

Brandon Carter of the Meudon Observatory in Paris has ad-
vanced a different observation-type of argument, based on an-
thropic considerations showing why the search for ETI looks to
be a bad bet. To outline his reasoning, consider the following
quantities:

tE = the time needed for evolution to produce an intelligent
species on Earth

to — the length of time that evolution can proceed on Earth

= the time during which the Sun remains in a state of
temperature and size amenable to life, i.e., the time the Sun
remains on the “main sequence”

tav = the average length of time needed to evolve an intelligent
species on an Earth-like planet

The Principle of Mediocrity says that tE sr tav, and that tav is
either much smaller or much greater than But both of these
expectations are contradicted by the fact that we observe t% ~

404

PARADIGMS LOST

t mg, which comes from assuming that the actual time needed to
evolve an intelligent species on Earth is about the same time
needed to evolve such an intelligent entity on a planet like
Earth. Furthermore, if there are many improbable steps on the
road to developing intelligent life, then we would expect to see tav
much, much greater than t^. Consequently, the observation that
*E ~ tm* is hard to justify beforehand, since a priori we would
have expected to find tE ;= tav. The implication that follows is
that tE is either much greater or much less than .

At this point Carter invokes the so-called Weak Anthropic
Principle, which for our purposes can be expressed by saying
that whatever we observe is biased by the presence of conditions
needed to ensure that we, as observers, exist to make the obser-
vation. We will give a much more detailed discussion of this
principle in the next chapter, but for now it suffices to note that
there is only a certain kind of universe that we could possibly
see: the kind in which the conditions are such as to allow our
own existence as astrophysicists who make the observations.
Using this kind of reasoning, Carter then argues that tE might
not come close to equaling tav. The fact that we observe tE ~
strongly implies that tav is much larger than tE, and that the
observed numerical coincidence tE ~ is due to the Weak An-
thropic Principle self-selection effect. Thus, Carter concludes
that tav is much larger than tE, which itself is about equal to t^.
Hence, the existence of ETI is highly unlikely since most Earth-
like planets will be destroyed by their star’s leaving the main
sequence long before intelligent life has a fighting chance to
emerge.

To cap off his claim, Carter derives a simple formula for how
long life on Earth can continue to evolve. Carter’s formula pre-
dicts that the biosphere on Earth can continue for at most an-
other 450 million years. This is a very short time indeed,
implying that the evolutionary window on Earth is already 90
percent of the way closed. Putting all his arguments together,
Carter concludes that there are no ETIs in the galaxy, and prob-
ably not anywhere else in the universe either.

The cases made by Hart, Carter, and Tipler for N = 1 are
representative of what we’ve termed above the observation cate-
gory of counter-ETI claims. Now let’s look at some of the fac-
torization-based claims that one or more of the terms in the
Drake equation must be negligible.

WHERE ARE THEY?

405

* * «

One of the most compelling cases of the factorization type has
been put forward by the philosopher Nicholas Rescher against
the likelihood of our being able to communicate with ETI, even
if it does exist. The standard pro-ETI argument for our being
able to engage in meaningful communication is that while ETI’s
social, political, and cultural systems may be radically different
from our own, its science is quite likely to be very nearly the
same. Rescher asks: Why should this necessarily follow? The
usual argument is the following anthropomorphic chain:

1. Common problems constrain common solutions.

2. ETI civilizations have in common with us the problem of cog-
nitive accommodation to a shared world.

3. Natural science as we know it is our solution to this problem.

4. Therefore, natural science is likely to be ETI’s solution, too.

Rescher notes that the obvious difficulty with this reasoning is
that ETI’s problems and ours are not the same, since the two
civilizations are literally worlds apart and have significantly
different environments and resources. To suppose a common
problem is to beg the question. Let’s consider for a moment what
it would take for ETI’s science to be the same as ours.

For ETI’s science to be functionally equivalent to ours and
hence form the basis for a meaningful exchange of information,
the following conditions would have to be the same:

• Formulation: The mathematics ETI uses must be the same as
ours. But there’s no reason why this must be so. For example,
it may use some kind of nonnumerical arithmetic.

• Orientation: It must be interested in the same sorts of prob-
lems that we are. But this may not be the case either, as ETI
may devote all its efforts to the social sciences, or might never
develop electromagnetic theory if its physical environment
doesn’t suggest it; e.g., ETI may live in on a murky world of
Stygian gloom where the main sensory inputs come from
sound rather than light.

• Conceptualization: It must have the same cognitive perspective
on Nature as we do. For instance, it’s not that seventeenth-
century biologists had something different to say about genes,
DNA, and the process of inheritance than we do; they had
nothing to say about these matters.

406

PARADIGMS LOST

Putting all these factors together, whatever ETI’s science is,
it’s going to be geared to ETI’s sensors, cultural heritage (which
determines what’s interesting), and environmental niche (which
determines what’s pragmatically useful). Sameness of the object
of contemplation does not guarantee sameness of the ideas about
it; e.g., primitive people regarded the Sun as a god, while we
think of the same object as a giant thermonuclear reactor. Note
that there is no argument here against the principle of the uni-
formity of Nature. The problem is that it’s different thought-
worlds that are at issue in the elaboration of science. Thus does
Rescher conclude that the quantity fc in the Drake equation
must be vanishingly small.

Michael Hart has contributed to the factorization school of ar-
gument as well, as noted earlier in discussion of his calculations
showing that the continuously habitable zone of a star is de-
pressingly narrow, suggesting that the term ne, the number of
planets that are suitable for life, is small. In addition, Hart has
argued that the term ft, involving the probability that life will
emerge, is also negligibly small. This line of reasoning is worth
examining in detail, especially as it occurs in several other bio-
logically based attacks on the Drake equation.

We have already seen from the Miller-type experiments that
many of the basic chemical building blocks of life can be formed
by natural chemical reactions in the primordial soup. However,
in order to have fi be large, it’s necessary to display a commonly
occurring mechanism by which these raw materials can form
themselves into self-replicating molecules of DNA. This is one of
the main points of attack on the Drake equation.

All earthly organisms have DNA strands that consist of a
chain of millions of individual nucleotides arranged in very spe-
cific ways. Any other ordering is usually useless, and may even
be fatal. This is why most mutations are deleterious and are
quickly weeded out of the gene pool. Hart supposes, for the sake
of argument, that on the early Earth there existed a “genesis
DNA,” which, when seeded into the primordial broth, acted as a
template around which other such strands formed, thereby giv-
ing the initial push needed to start evolution on its merry way.
Suppose that this prototype replicator needed only a sequence of
six hundred nucleotides in order to work, rather than the mil-
lions required by modern DNA. Further, imagine that only one
hundred of the six hundred positions on the strand have to be

WHERE ARE THEY?

407

occupied by a particular nucleotide element (A, G, C, or T), with
the remaining five hundred positions capable of being occupied
by any of the four bases. Then by random assembly of the 4
nucleotide bases, there are 4100 possible chains of 100 units.
After a few more calculations, Hart comes to the conclusion that
the chances of such a strand of genesis DNA spontaneously form-
ing are less than 1 in 1032. This number is inconceivably small,
implying that /, is essentially zero.

But if fi s; 0, what are we doing here? If the likelihood of a
particular event is negligible, how can we account for our own
existence? This Pact B obviously requires an explanation, par-
ticularly as it flies in the face of the cherished Principle of Medi-
ocrity. Hart’s ingenious answer to this dilemma is to point out
that according to modern cosmological thought, the universe is
not finite, but infinite! Therefore, even though the chances of
success of any single experiment are negligibly small, there will
be an infinite number of experiments, thus assuring many suc-
cesses; in fact, a gambler’s dream come true — an infinite number
of winners. Consequently, if we accept this line of reasoning,
there are an infinite number of planets where life has formed,
but these planets are so sparsely sprinkled throughout the uni-
verse that our chances of ever meeting with an ETI from one of
them are essentially zero.

Physicists and astronomers like Hart are not the only ones to
have made this kind of calculation. The eminent evolutionary bi-
ologists Ernst Mayr and George Gaylord Simpson have put
forth similar, but less quantitative, discussions of ETI using ex-
actly the same chain of reasoning. Mayr points to the convoluted
combination of seemingly chance circumstances that led from
the primordial slime to modern technological humanoids, noting
that the ancient Greek, Chinese, and Mayan civilizations were
created by individuals essentially indistinguishable from us
anatomically, yet they never developed a technological society.
Simpson makes what amount to the same kinds of arguments,
with both biologists concluding that money spent on SETI is
betting in a game whose odds are the most adverse in history.

A common thread running through all these biological and
sociological objections to the Drake equation is the assumption
that each slot on the nascent DNA strand has to be filled indepen-
dently. To illustrate, imagine you were set the task of construct-
ing a necklace consisting of one hundred beads, each of which is

408

PARADIGMS LOST

to be one of four colors: red, blue, green, or yellow. Further,
suppose that for aesthetic reasons the beads have to be arranged
in a very particular order, so that each of the hundred positions
must be occupied by a prespecified color. Now imagine you start
dipping your hand into a barrel full of beads of the various col-
ors and begin to assemble the necklace by placing the first bead
you get in position 1, the second in position 2, and so on. What
are your chances of getting the necklace put together properly in
one hundred trips to the barrel? Under the assumption that each
of your turns at the barrel has an equally likely chance of turn-
ing up any of the 4 colors, then the chances of selecting 100 col-
ors in exactly the right order are 1 in 4100, just the odds used by
Hart in his analysis of genesis DNA. In short, even if all the
people on Earth spent all their time trying, the odds are in-
finitesimally small that the necklace would ever be completed.
But complex systems in Nature just are not put together like
necklaces. Let’s see why.

The assembly scheme coming out of the independence assump-
tion implies that after a necklace one hundred beads long has
been assembled, we examine it bead by bead to see if every posi-
tion is occupied by the right color. If not, then the entire neck-
lace is torn apart and we start over from scratch. Nature works
in quite a different manner. On our trial necklace, even though
not all the positions are occupied by the right color, many will
be. In fact, there’s a 25 percent chance that any particular loca-
tion on the necklace will have a bead of the correct color. So if
the necklace as a whole isn’t perfect, we keep the part of it that
contains the right color in the right place and throw away only
those parts of it that don’t match up. What we might have after
the first round of such an experiment is something like the
string seen in Figure 6.15, where X represents a proper match
and O denotes a color mismatch.

0-0-0-0-0-0 • • • 0-0-0

FIGURE 6. 15. A trial necklace

For the next trial, all the pieces corresponding to X’s will be
kept and only those necklace fragments having mismatches will
be filled from the barrel. It’s easy to see that with this “ratchet-
ing effect” of keeping the subsystems that somehow “work,” the
entire necklace can be assembled in rather short order.

The above ratcheting principle forms the basis for Herbert

WHERE ARE THEY?

409

Simon’s Watchmaker Parable, illustrating the way in which
complex systems can be formed out of individual subsystems.
We briefly looked at this parable in Chapter Pour, the idea
being simply that it’s far quicker to form a complex system of
one hundred pieces from ten subsystems of ten pieces each than
it is to try to assemble a single system of one hundred compo-
nents. Computer experiments using this idea for the assembly of
genesis DNA have been made by the chemists Manfred Eigen
and Peter Schuster, as well as by the biologist Richard Dawkins,
all of whom come to the conclusion that formation of genesis
DNA using a dependent and directed, rather than an indepen-
dent and random, assembly of nucleotides from the primitive
components coming out of a Miller-type experiment is perfectly
feasible within a geological time frame.

The fly in the ointment is that in order for the ratcheting
principle to work, it’s necessary for the assembler of the neck-
lace to know what the necklace is supposed to look like. There
has to be a target design that all this shuffling-about of beads is
aiming at. Otherwise, there’s no way of telling whether a partic-
ular fragment should be kept or discarded. This is all very remi-
niscent of Douglas Hofstadter’s Jumbo computer program
described in the last chapter, which tries to do anagrams by a
directed assembly of individual letters. In that case, there are
very well understood and definite targets — recognizable words of
the English language. But if Nature is engaged in trying vari-
ous combinations of nucleotides to find a string that will self-
replicate, how can she decide whether a particular fragment is
or is not part of such a string before the entire string is assem-
bled? Unless a way out of this dilemma is found, one is thrown
back to the case considered by Hart, Mayr, and Simpson. At
present no one has any clear-cut idea of how to break through
this crucial bottleneck in the Drake equation, which serves as
our cue to conclude the arguments for the Defense and move on
to final summaries.

SUMMARY ARGUMENTS
Let’s first recall the precise question to be settled:

Is N, the number of civilizations within our galaxy with which we
are capable of communicating, greater than one?

410

PARADIGMS LOST

Note that I have highlighted the crucial point that our concern
is only with ETI civilizations within our own Milky Way Gal-
axy, and even then only with those ETIs with whom we can ex-
change meaningful information. Thus, the question as stated
must be answered negatively if the nearest ETI is in An-
dromeda, or if we encounter a clearly living but totally incom-
prehensible ETI like Lem’s sentient ocean in Solaris.

The various subgroups of the N > 1 position are displayed in
Table 6.4, along with representative members of the different
groups and a brief indication of the arguments they employ to
defend their positions. I hasten to point out that in some cases I
have made use of a bit of literary license to assign certain in-
dividuals to particular groups, since their writings are not abso-
lutely explicit as to precisely where they stand regarding the
magnitude of N. Nevertheless, on the basis of what they have
written I feel the assignments of Table 6.4 are close enough for
government work, and certainly acceptable for our purposes
here. Following the Prosecution’s summary, Table 6.5 displays
the various counterclaims offered by the Defense.

So there it is: the usual collection of eminent scientists stri-
dently arguing mutually exclusive positions. After the smoke
clears, the situation seems ultimately to turn upon an act of

N > 1 : ETI EXISTS!

PROMOTER

ARGUMENT

(N is large)

Sagan, Morrison Principle of Mediocrity

Dyson

Papagiannis

(N is small or large)

comets or Dyson spheres

asteroid belt

Drake

(N is moderate)

travel/colonization too expensive

Rood

Bracewell

(agnostic)

Drake equation
von Neumann probes

TABLE 6.4

Summary arguments for the Prosecution

WHERE ARE THEY?

411

N = l: ETI DOES NOT EXIST!

PROMOTER

ARGUMENT

Hart

no colonization; fe small

Tipler

absence of von Neumann probes

Mayr, Simpson

f,, f, L small

Trefil

no colonization

Carter

Anthropic Principle

Rescher

otherworldly science

TABLE 6.5. Summary arguments for the Defense

faith — just as Frank Tipler claimed. In fact, the situation is
strikingly similar to the famed psychologist Carl Jung’s analysis
of alchemy: “When facts are few, speculations are most likely to
represent individual psychologies.” If you’re a believer in the
Principle of Mediocrity, then it’s inconceivable that N could be
small; on the other hand, if you have faith in the irresistible
urge of all living things to seek out new worlds, then you have to
feel that N = 1 and we are that numero uno. As to my own
brand of spiritual firewater, read on.

BRINGING IN THE VERDICT

To the ETI question as stated, I vote for acquittal, thus sup-
porting the Defense argument that N = 1. Oddly enough, while
most of the Defense arguments center about the Fermi Paradox
and the issue of colonization, it is not this line of reasoning that
leads me to side with the Defense. Nor is it a rejection of the
Principle of Mediocrity forming the heart of the Prosecution’s
case. Rather, my view is that ETI may very well and probably
does exist, even somewhere in the Milky Way. However, what I
find difficult to swallow is the implicit corollary of the Principle
of Mediocrity that if ETI is around, we will be able not only to
recognize it, but even to enter into some sort of meaningful dia-
logue. In this regard I find the arguments put on the table by
Rescher difficult to rebut. And in view of these arguments I
think the issue is not that ETI science may be more advanced
than ours. Rather the issue is that the likelihood is essentially

412

PARADIGMS LOST

zero that they will be doing our sort of science at all. So there
may well be intelligent extraterrestrial civilizations out there,
but the chances are negligible that we’ll ever contact one doing
“our kind” of science. Thus it’s at the communication level that
I draw the line, and since communication is an integral part of
the judge’s charge to the jury, I have little recourse other than
to conclude that N = 1. Hence, my vote for the Defense.

While it’s not part of my argument for JV = 1, what can we
expect if SETI is actually successful and a signal from the deep
is received? The conventional wisdom of the pro-ETI crowd al-
ways emphasizes how the receipt of such a signal will pro-
foundly change our concept of ourselves. Just what could this
actually mean? At one end of the scale, if the signal shows that
the entire universe is run by a band of angelic swans from
61 Cygni who control every aspect of our lives, then such a dis-
covery would indeed have profound implications for our notion
of self. If, on the other hand, the signal shows that there is a
“second Earth” out there where ETIs worry about stock market
crashes, go on vacations to “Hawaii,” and play baseball, then
the message would probably result in a vast, almost unbelievable
disappointment, but would surely not influence our self-concept
in the slightest. So just what are the advantages of detecting a
signal, other than of course satisfying our curiosity?

The benefits of a message from the stars will ultimately de-
pend upon whether the ETI civilization is sufficiently close to
ours for a meaningful transmission of useful information. If the
civilization is totally alien, then there will really be nothing to
learn from the signal since we will have nothing at all in common
with it. After all, what could we have to say to members of a
civilized, technological species inhabiting the surface of a neu-
tron star, living out their lives in a fraction of a second (by our
clocks)? Robert L. Forward thought there was something to say
in The Dragon’s Egg, but interesting as his arguments are, I’m
skeptical. In fact, a message from such an alien may not even be
decipherable, in much the same way that the mysterious Voy-
nich manuscript here on Earth has defied all attempts to ferret
out its meaning.

If the ETIs are close enough to us for some sort of meaning-
ful exchange of information to take place, unless their message
was specifically tailored to a culture like ours at our point of
development, their science would probably be no more compre-

WHERE ARE THEY?

413

hensible to us than the wiring diagram of an IBM PC is to an
aboriginal tribesman. And when it comes to political, cultural,
and ethical information, the signal may suggest practices or sys-
tems that we would find immoral or just plain unworkable, e.g.,
rationing of children or abolition of money.

To conclude on a somewhat sober note, I find the supposed
benefits of SETI to be vastly oversold by the pro-ETI en-
thusiasts, principally because I think that even if ETIs are out
there, we’ll either never know it or we’ll never get any real bene-
fit from it, simply because they are truly and fundamentally
alien. Consequently, when dividing up the research-dollar pie, it
seems to me a poor investment to put too much faith, hope, or
money into SETI on the grounds that the expected return from
a message will enable us to amortize the investment easily, even
if the signal comes hundreds or thousands of years from now.
On the other hand, a few million dollars a year is petty cash
from the NSF and NASA budgets, so why not spend a little
money now and then on looking for the pot of gold at the rain-
bow’s end? After all, curiosity is a wondrous thing, and it’s hard
not to wonder when you look up at a clear night’s sky, “Where
are they?” We’ll never know if we don’t look. And looking and
hoping are what science is all about.

HOW REAL IS THE
"REAL WORLD"?

CLAIM

THERE EXISTS NO OBJECTIVE REALITY
INDEPENDENT OF AN OBSERVER

BUILDING THE STAGE

In his comedy As You Like It, Shakespeare makes the well-
known statement that “all the world’s a stage, and all the men
and women merely players.” This Shakespearean remark con-
jures up the commonsense, everyday view of physical reality:
The universe of material objects — chairs, cars, trees, atoms — ex-
ists independently of us, just as the theater and its stage exist
independently of the actors and their audience. This image of an
impersonal, aloof cosmos was engraved onto the scientific con-
sciousness by the authority of Newton and his idea of events

HOW REAL IS THE ''REAL WORLD''?

415

that unfold in an arena of absolute space and time. Since this
idea forms the framework upon which our story in this chapter
is draped, let’s quickly review the essential components of the
world according to Newton.

The essence of the “objectivist” position, nowadays termed
naive realism, is that the world consists of a collection of inde-
pendently existing “things” that are simply “out there” whether
we observe them or not. To be more specific, we can identify the
principal components of this ontology as follows:

• There exist identifiable things that possess intrinsic attributes.

• It is not necessary for these things actually to be observed in
order for them to exist.

• We as observer/participants are part of this reality, but think
of it as being independent of us and as existing both before
and after ourselves.

• The observers have predetermined roles to act out within the
framework of this reality.

Einstein himself pithily summarized the core of this taken-for-
granted reality when he remarked to Pascual Jordan, “Do you
really think that the Moon exists only when you look at it?” In
his view, it was the job of science to go beyond mere surface
appearances and to describe and understand the nature of this
objective, independent-of -human-affairs, rock-bottom kind of
physical reality.

The Newtonian picture is by now so deeply ingrained in our
ways of thinking about life, the world, and the universe that it’s
hard to imagine anyone doubting it. And indeed few did until
early in this century, when the relativity and quantum theorists
recognized that Shakespeare and Newton had always been living
behind the facade of a Potemkin village, at least when it came to
dealing with the very small, the very large, and the very fast.
But the world at large, including most practicing physicists, was
happy to accept the tacit assumption that these non-Newtonian
effects really count only in the microworld of the atom or the
macroworld of distant galaxies. And it’s been only in the past
decade or so that the reality crisis of the physicists has spilled
over into the realm of everyday life, with accounts in both the
popular and the New Age press unveiling for the general public
such seemingly romantic notions as observer-created realities
and the intertwining of modern physics and Eastern mysti-

416

PARADIGMS LOST

cism — with the blessings of at least some renegade physicists, no
less! To get a glimpse of what’s involved in this wholesale re-
vamping of our concepts of physical reality, there’s no better
place to start than with the familiar parlor game of twenty ques-
tions.

A common form of the twenty-questions game involves a
group of people who send one of their number out of the room to
act as the questioner. The group then decides upon a target word
and the banished party is asked to return. It is then the task of
the questioner to identify the target word using at most twenty
questions, such as “Is it alive?” or “Is it liquid?” The winner of
the game is that questioner who identifies the target word using
the smallest number of questions, under the stringent condition
of having only one chance at actually guessing what the word is.

The physicist J. A. Wheeler likes to tell of the time he played
an interesting variant of the game following a dinner party at
the home of physicist Lothar Nordheim. According to Wheeler,
he was sent from the room for what seemed an inordinate length
of time. Returning to the room, he saw a smile on everyone’s
face — a sure sign that some sort of mischief was afoot. He then
started his questioning with the customary sweeping queries:
“Is it animal?” No. “Is it mineral?” No. “Is it alive?” No. But
as the questioning went on, Wheeler noted that the answers were
slower and slower in coming, with the person being questioned
thinking for a long time before responding with a simple yes or
no. Finally Wheeler felt he had narrowed the possibilities down
to the point where he was ready to take the plunge. “Is the word
‘cloud’?” he asked. At which point everyone broke out laughing
and told him he was correct. It seemed that while he’d been out
of the room the others had agreed that they would not select any
word, but rather would let some word emerge as a consequence
of Wheeler’s questioning. The agreement was that the parties
being questioned could respond with either a yes or a no, the
only constraint being that whichever response they gave, they
would have to have a definite word in mind that would be con-
sistent with all the preceding responses. So the game was at least
as difficult for the others as it was for Wheeler!

The point Wheeler makes when recounting his twenty-ques-
tions story is that the game serves as a metaphor for two com-
peting versions of what constitutes physical reality. Let’s call

HOW REAL IS THE "REAL WORLD''?

417

them objective and contextual reality. Objective reality corre-
sponds to the standard form of the game in which the word is
preselected. This is just our old friend Newtonian reality again.
The things (words) of this world exist and have real properties
independent of human observers or measuring devices. Wheel-
er’s game corresponds to a contextual reality, and involves a
world that is literally created by the way in which it is probed
by the observer. Just as there was no definite word but only po-
tential words when Wheeler (the observer) entered the room, no
stage is out there waiting for us to step forward and read our
lines either. This situation calls to mind Gertrude Stein’s with-
ering assessment of Oakland: “There’s no ‘there’ there.” Actu-
ally, there are only potential “theres,” and the stage of reality is
constructed in real time as we proceed to act out our roles as
observer/ participants .

So is Wheeler’s word really there or isn’t it? Is there an hon-
est-to-god objective reality underlying the surface appearance of
things? Or is it necessary to introduce some kind of observer as
the creator/constructor of what we think of as being “real”?
Shakespeare, Newton, and my barber say yes, the world really is
“there”; the modern quantum physicist tells us maybe not. To
see why, as well as to understand the many senses in which
Wheeler’s word and our world might not really be out there at
all, we must set out on an all-too-brief tour of a few prominent
landmarks in the wonderfully weird world of the quantum.

GHOSTS IN THE ATOM

Newton’s world is a world of particles and forces. One might
think of it as a world composed of little billiard balls, each char-
acterized at any given moment by three attributes: a mass, a po-
sition in space, and a speed of movement in some spatial
direction (technically, a velocity). The mass is what we call a
static attribute, since its value doesn’t change during the course
of time. The position and velocity are examples of dynamic attrib-
utes. Everything that happens in Newton’s world happens as a
result of these little balls flying around, colliding, combining,
and breaking apart according to forces acting upon them from
the outside. The formula for these interactions has been en-
shrined in the physicist’s lexicon as Newton’s Second Law, and

418

PARADIGMS LOST

is expressed in the form a = F/m, i.e., the acceleration of a par-
ticle (the rate of change of its velocity) equals the force imposed
upon it divided by the particle’s mass. As to the nature of these
mysterious imposed forces, Newton, cagey as ever, evaded the
issue entirely with his classic disclaimer hypotheses non jingo (I
make no hypotheses).

In this Newtonian kind of universe, everything is unbelieva-
bly tidy and orderly. As soon as the imposed forces are specified,
together with the initial position and velocity of each particle,
events unfold with metronomic regularity upon a preexisting
stage of space and time. Implicit in this rosy clockwork world is
the assumption that the attributes of the particles are present at
each moment, quite independently of whether or not there is a
voyeur on the scene taking a quick peek at them with some kind
of measuring device. The unchallenged success of this New-
tonian picture in predicting phenomena of concern in the eigh-
teenth and nineteenth centuries, coupled with the close
agreement between the billiard ball metaphor and everyday com-
mon sense, led to a kind of “soft brainwashing” of both the
scientific community and the general public. The prevalent belief
of those times was that

Newton’s universe = The real universe

The first cracks in the Newtonian facade came with the Spe-
cial Theory of Relativity, in which Einstein showed that the
playing field of space and time could not be as clear cut as New-
ton thought. In fact, the Special Theory showed that the only
kind of reality consistent with observational evidence was one in
which space and time were not considered as separate entities at
all, but as a single indivisible unit — spacetime. Furthermore, the
Special Theory asserted that the separation between two given
events observed in the new playing field of spacetime might be
seen as positive by one observer, negative by another. In short, the
two different observers could be seeing two quite different
“realities,” making it impossible for them to agree upon the an-
swer to even such a seemingly simple question as which of the
two events preceded the other.

With the introduction of the idea that there is no such thing
as an objective, observer-independent event, at least insofar as
describing its location in space and time, Einstein showed that
there was something fishy about the kind of reality that Newton

HOW REAL IS THE ''REAL WORLD''?

419

had in mind. However, as has often pointed out in the past, Ein-
stein’s work was in many ways the last gasp of the Newtonian
world, as neither the Special nor the General Theory of Relativ-
ity had much to say about the material objects themselves. On
matters pertaining to the static and dynamic attributes of New-
ton’s particles — e.g., mass, electric charge, velocity, spin — rela-
tivity theory is silent or, more accurately, tacitly accepts the
Newtonian precepts hook, line, and sinker. Instead Einstein’s
theories focus upon the other half of the Newtonian doublet, the
unexplained forces (particularly gravity), in effect centering at-
tention on the nature of the playing field on which the particles
act out their predetermined Newtonian destinies. Especially in
the General Theory, which is “nothing more” than a general the-
ory of gravitation, Einstein showed that the playing field itself
is in some way created by the particles, which are then told how
to move by the topography of the terrain they generate. So
rather than having an independent reality of its own, the play-
ing field exists in a kind of symbiosis with the players. This is a
queer enough notion in its own right, at high variance with
human perceptions generated and sharpened by the events and
vicissitudes of everyday life. But when it comes to weirdness,
you ain’t seen nothin’ yet.

At about the same time Einstein was slaving away at the
Swiss Patent Office in Bern putting the finishing touches on the
Special Theory, Max Planck in Berlin was working the other
side of the Newtonian street with his discovery of the quantized
nature of the radiation given off by a hot object. This work
showed that some of the basic quantities of physics, like energy
and angular momentum, come in minimum-sized “chunks.” In
particular, Planck demonstrated that light of any energy comes
in such chunks whose size depends upon the frequency of the
light, i.e., its color. The implications of this work drove the last
nail into the coffin of Newtonian reality, serving as the impetus
for what today J. A. Wheeler terms recognition physics: the study
of why there are such things as time and space and dimensional-
ity at all. Just as it’s impossible to say you’ve really visited
America without seeing the Statue of Liberty, the Grand Can-
yon, and the Golden Gate Bridge, it’s equally impossible to talk
about the “reality of reality” without visiting a few of the
sights in the land ruled by the iron hand of the quantum. So
let’s start the tour.

* * *

420

PARADIGMS LOST

To understand the profound implications of quantum theory
for describing the way the world really is, there’s no better place
to start than with three versions of the traditional double-slit
experiment. The experimental setup includes a projector, which
produces three different types of material objects upon com-
mand: bullets, water waves, and electrons. For any given run of
the experiment, only one of these types is produced. Whichever
type of object is chosen, the device projects it toward a screen
containing two slits (or gaps), either or both of which may be
open. Behind the screen sits a line of detectors capable of regis-
tering the appearance or absence of the projected objects after
their passage through the screen. Now let’s run a few experi-
ments.

First of all, suppose the projector is set to produce a stream
of bullets. Figure 7.1 shows the results of three such experi-
ments: with slit 1 open, with slit 2 open, and with both slits open.
We’ll call the number of bullets reaching the detectors in each
case P, , P2 , and P,2 , respectively. In the figure, the bullets
passing through slit 1 are shown as white-centered circles, while
those passing through slit 2 are depicted as solid black circles.
It’s important to notice here that the number of bullets reaching
each detector when both slits are open is just the sum of the
numbers obtained when only one or the other of the slits is open.
This is exactly the result we would expect to obtain from the
classical view of bullets as individual particles going about their
appointed rounds. Now let’s change the projector setting from
bullets to water waves and see what happens.

In Figure 7.2 the projector sends water waves instead of bul-
lets through the slits (which we now might envision as gaps in a
jetty) and on to the line of detectors. In this case the detectors
can be thought of as floating buoys whose bobbing up and down
measures the height (energy level) of the waves passing beneath
them. The symbols I, , I2, and I,2 denote the situations in which
gaps 1, 2, and 1 and 2 are open, respectively.

A crucial point to notice here is that when either gap 1 or gap 2
is open, the detection pattern is similar to that seen when pro-
jecting bullets. But when both gaps are open, the patterns di-
verge dramatically. This divergence is the result of the
phenomenon of wave interference in which two waves can inter-
act to form a new composite wave, either by reinforcing each
other through constructive interference, or by neutralizing each
other by means of destructive interference at places where peaks

HOW REAL IS THE ''REAL WORLD''?

421

□

Detect

Boxes

^12

DqoII

&:::

ploio

koo:

»:::

kco:

»:::

c*oo#o::

com:

»:::

ioicoo::

k oo:

m::

fpiooio*-

koo:

&*::

SoicO *5"

cooo:

&»::

D#C0*5f:

coco:

K*D##CQ

ko::

p*o*o

coo::

Omomo

co:::

khq*::

co:::

mz:

|BB*“

Slit 2 open

Slits 1 and 2

Slit 1 closed

open

Number of bullets in each box after a fixed time
FIGURE 7.1. The double-slit experiment with bullets

floating t

buoys v

Wave intensity at each buoy)

FIGURE 7.2. The double-slit experiment with water waves

in one wave encounter troughs in the other. The basic idea,
which is important for our later discussions, is depicted in Fig-
ure 7.3. Now let’s again flip the projector switch and this time
produce electrons instead of water waves.

In this experiment we can regard the slits as two holes in a

422

PARADIGMS LOST

thin metal plate and the line of detectors as elements on a phos-
phor screen. This is just like your TY set, where the electrons
are shot from a gun at the back of the picture tube and are
focused by an electrostatic lens to strike the picture screen at
just the right spots. The results are shown in Figure 7.4.

When we open only slit 1 or slit 2, we get the same pattern
seen in Figure 7.1 (the bullets). In Figure 7.4, the open-centered
circles represent electrons passing through slit 1, and the solid
black circles those that pass through slit 2. The surprise occurs
when we open both slits. Column PJ2 shows the same interfer-
ence pattern we saw in the experiment using water waves, hence

HOW REAL IS THE "REAL WORLD"?

423

Number of electrons arriving at each detector
(in a fixed time)

FIGURE 7.4. The double-slit experiment with electrons

it, too, necessarily involves some kind of wave motion with inter-
ference effects. In this case, however, Pl2 is not the sum of col-
umns P, and P2 , so we can’t say which of the two slits any
particular electron went through. This fundamental lack of
knowledge is indicated in the figure by representing the elec-
trons as half black and half white. It’s crucial to note here that
the electrons still arrive at the phosphor screen as individual
particles, i.e., as “bullets.” It’s just that their pattern of arrival
makes it look like they collectively obey some sort of wavelike
law of motion, making it impossible to assign a given slit to a
given electron. Thus we arrive at:

THE MYSTERY OF THE QUANTUM WORLD

How can electrons possess the attributes of both particles and waves,
yet behave like neither ?

With the discovery that the fundamental particles of New-
ton’s world — the constituents of the atom — have such surprising
and contradictory behavior, the last thread of support for classi-
cal Newtonian reality was cut away, confronting physicists with
the task of describing and explaining a wondrously bizarre new

424

PARADIGMS LOST

world. Comparatively speaking, the description part of this dou-
ble-sided chore turned out to be easy. But the explanation of
what the description means divides the community of physicists
and philosophers to this day. So let’s start with the easy part,
and work our way into the conundrums of present-day thinking
as to the true nature of quantum reality.

From our discussions in Chapter One, the reader will recall
that for a theoretical scientist, to describe some phenomenon
means to construct a mathematical representation or model of
the phenomenon that takes into account all the aspects of inter-
est about it. In the case of quantum objects like the electron, this
means that we need to find some kind of mathematical structure
that encompasses static attributes like charge and mass, as well
as dynamic attributes like position, momentum (mass times ve-
locity), direction of spin, and so forth. Furthermore, our mathe-
matical structure must reflect the strange behavior discussed
above, in which the electron shows the characteristics of both a
particle and a wave without actually being either. The solution
to this Quantum Description Problem is a tall order, yet one
that was filled rather rapidly in no less than three different ways
by Werner Heisenberg, Erwin Schrodinger, and Paul Dirac
about sixty years ago. As it turned out, these seemingly different
mathematical descriptions all ended up being mathematically
equivalent. So I’ll content myself here with just a brief sketch of
one of the solutions (Schrodinger’s), since even today it forms
the main weapon in the working physicist’s mathematical arse-
nal for dealing with quantum phenomena.

The heart of Schrodinger’s scheme is to represent the “state”
of a quantum entity like an electron at any moment by a mathe-
matical gadget displaying wavelike behavior. What this means is
that such an object can show the type of interference phenomena
associated with waves when it interacts with other such objects.
As a key component in his solution to the Quantum Description
Problem, Schrodinger derived an equation that tells us how the
state of the object changes at each point in space over the course
of time. In this solution the state somehow encapsulates all the
dynamic attributes that the object can possess. So to calculate
the chances of any particular value of any one of these attrib-
utes turning up at any particular moment if a measurement is
actually made, Schrodinger argued that we must perform some
additional mathematical operations on the state to extract the

HOW REAL IS THE ''REAL WORLD''?

425

desired likelihoods. While the technical details are out of bounds
here, the basic idea is rather simple to describe.

To fix a specific situation, suppose we have an atom existing in
an excited energy state. Such an atom gives up energy by throw-
ing off an electron, just as in the Planck experiments discussed
earlier. Quantum mechanics represents this event as a wave
function that spreads out from the atom in an ever-widening
spherical wavefront. This is exactly like dropping a stone into a
pond and watching the ripples of water move away from the
stone. The amplitude of this spreading wave function at a point
in space and time gives the probability of finding the electron at
that location at that point in time. Now suppose that the electron
eventually runs into a silver atom in a piece of photographic
film. When it hits the film, the electron gives up its energy and
leaves a black spot on the film. At that precise moment the elec-
tron’s wave function “collapses” in a way that is reminiscent of
the breaking of a soap bubble. The wave function disappears
from all of space except the region of the struck silver atom.
Since the electron has given up all of its energy to the silver
atom, it has no probability of existing elsewhere. The wave func-
tion vanishes or, more properly, becomes a “spike” at the loca-
tion of the silver atom in space at the time of the collision.
Keeping this concrete situation in mind, let’s now turn to the
general situation described by Schrodinger.

Suppose the quantity W(x , t ) represents the wavy state of
the particle at time f in a spatial region described by the quan-
tity x . Further, imagine that A represents the attribute we want
to know about, e.g., the particle’s position. Schrodinger showed
that each such attribute can be associated with its own charac-
teristic family of waveforms. A sample of some of these wave-
form families is shown in Figure 7.5. The figure, incidentally,
shows why some attributes can assume only quantized values
while others can take on a continuous spectrum of values. Wave-
form families, like the spherical harmonics, are constrained in
the kinds of vibrations they can display by the geometrical re-
gion of their action. Thus, such waveform families can vibrate
only at certain resonant frequencies, while all other frequencies
are physically inaccessible. Unconstrained families like the sine
waves can vibrate at any frequency whatsoever, hence the attri-
bute corresponding to such a family can assume a continuum of
values.

Suppose the waveform family corresponding to the attribute

426

PARADIGMS LOST

WAVEFORM-ATTRIBUTE DICTIONARY

Waveform

Piano wave

Attribute

Momentum

Spin

Unnamed attribute

FIGURE 7.5 -4 waveform-attribute dictionary

HOW REAL IS THE ''REAL WORLD''?

427

A (such as an electron’s position) is denoted by »,(*, t ),
w2{x , t ), w,(x, t), . . . Part of Schrodinger’s genius was to see
how to associate each member of such a family with one of the
possible values that the attribute A can assume — once it’s actu-
ally measured. In general, each such family has an infinite num-
ber of members; hence, usually an attribute A has the possibility
of taking on any one of an infinite number of possible values.
Again, whether the number is quantized or not depends solely
upon the waveform family associated with A . For example, if A
is the position of an electron within a closed box, then at any
particular moment the electron might be found at any one of the
infinite number of spatial locations within the box. Appealing to
general mathematical results, it can be shown that the state
W(x, t ) may be uniquely decomposed in terms of the waveform
family associated with the attribute A . This means we can find a
set of numbers c„ c2, c3, . . . such that

W(x, t ) = c,w,(x, t ) + c2w2(x, t ) -(- c,iv2(x, t) + . . .

where {w2(x, t), w2(x, t), . . . ) is the waveform family corre-
sponding to the attribute in question, e.g., position.

A good way of thinking about this decomposition process is to
recall the grade-school science experiment in which your teacher
shone ordinary white light through a prism and a rainbow came
out the other side. In the quantum case, the object’s wave func-
tion W(x, t) corresponds to the white light; the waveform fam-
ily [w{(x, Oii1 = 1, 2, . . . , to the various colors of the rainbow.
In this metaphor, each particular attribute of interest about the
object of study corresponds to a different prism through which
we can view the wave f unctiorr.^f course each type of prism will
break up the wave function into its own particular “rainbow,”
so we’ll get a different waveform family (wt (x, Oj and a differ-
ent set of numbers jc, j in our decomposition depending upon
which attribute (prism) we’re using to separate W. Now, how
does the above decomposition allow us to get a handle on the
spread of values that the attribute A might take on?

Recall that in the decomposition there is a unique number c,
associated with each family member w{(x, t). Furthermore, by
Schrodinger’s scheme there is a way to pair up wt(x , t ) with the
ith value that the attribute A might conceivably assume when
measured. Then Schrodinger’s rule for the likelihood of dynamic

428

PARADIGMS LOST

attribute A taking on its tth possible value is simplicity itself:
Just square the quantity c,. That’s all there is to it. Just multi-
ply the number c, by itself and the result will be the probability
that, when measured, the value of the attribute A will be found
to be its tth possible value. (Technically, the number c, is a com-
plex number, not real. Thus, we should use the complex modulus
rather than c\. For details, see the To Dig Deeper section.) Of
course, the specific numerical value seen when we do measure A
will be conditioned by the precise correspondence between the
waveform w,(a:, t ) and the set of theoretically possible values
that A can take on. But the underlying idea of how to calculate
the dispersion of possible experimental outcomes is, I think,
clear and straightforward.

Since the idea is so central to all of quantum theory, let’s reca-
pitulate the steps in Schrodinger’s solution to the quantum de-
scription problem.

QUANTUM DESCRIPTION ACCORDING
TO SCHRODINCER

1. Calculate the wave function W(x, t ) for the given experi-
mental situation from the Schrodinger equation.

2. Decide which attribute A you wish to measure.

3. Look up the waveform family (w/,(a:, t)},i = 1, 2, . . . corre-
sponding to A in the waveform-attribute dictionary.

4. Decompose the wave function in terms of the appropriate
waveform family as W{x, t ) = c,w, (x, t) + c2w2(x, t ) +
c3w, (x,t) + . . .

5. Compute the probability that A will assume its ith possible
value by squaring the number c,.

Let’s pause here for a moment and reflect upon the dramatic
difference between the above prescription for describing a quan-
tum object such as an electron, and Newton’s procedure for de-
scribing a classical particle like a bullet. For the bullet, the
Newtonian description regards the state as being the actual posi-
tion and momentum of the bullet at any instant of time; for
Schrodinger, the state is the wave function, which measures only
the likelihood that the particle (an electron, say) has a certain
position (or momentum) at a given time. Conceptually and oth-
erwise, these are radically different views of the “reality” of the
particles’ attributes. In the Newtonian case there’s no question

HOW REAL IS THE ''REAL WORLD''?

429

but that the position and momentum are innate attributes of the
bullet existing at all times. For the electron, the Schrodinger de-
scription is silent on the matter of the innateness of these attrib-
utes, and only gives a prescription for how to compute the
likelihood of an attribute’s taking on a given value when a mea-
surement is actually performed. Note that this is true even
though the traditional quantum view outlined above has tacitly
reinstated the Newtonian vision of absolute and separate space
and time. To incorporate Einsteinian spacetime into a quantum
description would take us right up to the forefront of contempo-
rary research on quantum gravity, far beyond where we either
can or need to go in an elementary account of this type.

After all is said and done, we come to see that the Newtonian
state of the particle (the position and momentum) has the ap-
pearance of something substantial and agrees with everyday
common sense. On the other hand, the quantum state (the wave
function W) appears as a physical fiction, a mere wave of proba-
bility, taking on a tangible quality only when a measurement is
actually made. Yet it would appear that this mathematical wave
is the very thing that’s needed in order to build a description
that is in harmony with what’s actually seen in the laboratory.
And there’s no sweeping the dirt under the rug either, since the
quantum description is the undisputed king of all theories of
physical phenomena, having been tested thousands of times in
laboratories and research centers around the world and never
yet failing to be in accord with what our instruments report.
Nevertheless, to physicists of a philosophical inclination and to
philosophers of a physical bent, the whole quantum business is
shrouded in an aura of mystery when it comes down to what it
all really means. This cloud of philosophical and physical uncer-
tainty hangs like a mist around the peaks of two sacred moun-
taintops on the quantum horizon: the Quantum Measurement
Problem and the Quantum Interpretation Problem. So as the
next stop on our package tour, let’s take a longer look at these
two so-far-unscaled peaks.

MEASUREMENT TO MEANING

When I first encountered the weirdness of the quantum world as
a student too many years ago, one of my first thoughts was

430

PARADIGMS LOST

“How could it be like that?” Little did I realize at the time that
my futile plea for an explanation of just what was going on had
already been answered by the late physicist, educator, and gen-
eral bon vivant Richard P. Feynman when he remarked:

I think it is safe to say that no one understands quantum mechan-
ics. Do not keep saying to yourself, if you can possibly avoid it,
“but how can it be like that?” because you will go down the drain
into a blind alley from which nobody has yet escaped. Nobody
knows how it can be like that.

I think that I left the pursuit of physics at just about the time I
came across this remark.

To my way of thinking then (and now), the Schrodinger solu-
tion was no solution at all, just a set of formulas and mathemati-
cal tricks for predicting the results of experiments. Not being of
a very practical orientation even then, I didn’t think this was
nearly enough. Somehow I thought physics was going to talk
about the world of reality, but what I found was merely a dis-
cussion of the world of phenomena. Only later, after abandoning
both the world of reality and the world of phenomena for the
otherworldly universe of mathematics, did I come to see more
clearly that perhaps the two worlds of reality and phenomena
could be brought into contact after all. Since the link between
them lies in the act of observation, i.e., measurement, the first
step in this rapprochement necessarily has to be a deeper under-
standing of what’s so special about the nature of measurement,
and why it seems to play such a distinguished role in the consid-
eration of quantum processes.

All solutions of the Quantum Description Problem, Schro-
dinger’s or otherwise, share a common feature: Prior to a mea-
surement, the quantum object is described only by a wavelike
quantity specifying the relative likelihood of an attribute’s tak-
ing on one or another of its potential values when actually mea-
sured. As we saw in the Schrodinger scheme, these likelihoods
are given by a set of numbers that, taken together, form a prob-
ability distribution for the outcome of an observation made on
the object. Of course, once the measurement is actually taken all
uncertainty fades away, since one of the possible values of the
attribute has been singled out by the measuring device. To ham-
mer home the point, suppose the attribute of interest is A and

HOW REAL IS THE ''REAL WORLD''?

431

that A can theoretically take on N possible values in a given
experimental situation. Let’s label these values vu v2, , v#.

Note that each of these values is just a symbol that might be
physically displayed as a pointer position on a dial, the number
of clicks from a counter, or some other form of output produced
by the measuring device once we actually do the experiment. By
the Schrodinger procedure discussed above, associated with each
such value v, is another number c J, the likelihood that the exper-
imental outcome will show the value r/,-, i = 1, 2, . . . , N. Thus
before the measurement is made we have the situation depicted
in Table 7.1.

Now suppose the measurement is made and it turns out that
the resulting value for the attribute in question is, say, v2 ■ Then
after the measurement the situation is that given in Table 7.2.
Thus after the measurement is taken, the a priori set of likeli-
hoods • • • >cw) has “collapsed” into the degenerate set

(0, 1, . . . , Oj in which every element is zero except the second,
corresponding to the actual outcome, which now has likelihood
equal to one, i.e., complete certainty.

As a particularly simple concrete illustration of the foregoing
situation, consider a spinning electron. We can think of this as a
basketball spinning on the end of some Harlem Globetrotter’s
finger, with the finger corresponding to the axis of spin for the
ball. Suppose a fixed direction in space is prescribed, and that
the attribute we’re concerned about is the component of the elec-
tron’s spin in that direction. The axis along which the electron is
spinning then either points in the direction in question or points
in the opposite direction. For the sake of definiteness, let’s call
the attribute value in the first case UP and in the second case
call it DOWN; i.e., for this experiment N — 2, v, = UP, and
v2 = DOWN. Incidentally, this example illustrates the point
that the “values” of an attribute don’t always have to be
thought of as numbers. They just need to be distinguishable la-
bels like UP and DOWN, characterizing different possible out-
comes of measurement. If we have no special information about
the electron, then its spin axis before measurement is equally
likely to be pointing in any direction. Consequently, it’s reason-
able to assume that the two possible outcomes are equally likely,
i.e., c i = c | = |. As soon as we actually measure the electron’s
spin, we find out in which direction its spin axis is pointing,
with the consequence that the a priori likelihood set [c\ =

432

PARADIGMS LOST

possible experimental outcomes

• vn

likelihood of outcome

r 2 r 2

C1 c2 *

• cjr

TABLE 7.1. The situation before making a measurement

possible experimental outcomes

V2 ■ •

• VN

likelihood of outcome

0 1 . .

. 0

TABLE 7.2. The situation after the measurement

c 2=2! collapses to either JO, lj if the axis points “DOWN” or
to Jl, 0) if it points “UP.”

In terms of the above experiment, we are now in a position to
state the essential features of the Quantum Measurement Prob-
lem. But before doing so, let’s pause to clarify one dangling
loose end: What is a measuring device anyway? In the commu-
nity of quantum theorists even this seemingly simple question is
unsettled. Some say a measuring device is any instrument capa-
ble of leaving a permanent record. According to this view, which
seems to concur with most people’s sense of what’s right and
proper, things like Geiger counters, meter sticks, and photo-
graphic plates all constitute valid measuring devices. But others
claim that the only kind of measuring device capable of collaps-
ing the quantum probability set (or, equivalently, the wave func-
tion) is consciousness; i.e., the observation has to enter a
conscious mind before the magical collapse can occur. And even
in this stringent view, it’s still unclear whether any conscious
mind will do, or whether only the kind of consciousness dis-
played by Homo sapiens will suffice. Can the probabilities be col-
lapsed by your family dog? By the roaches in the kitchen? By an
amoeba? No one is saying for sure. So for the moment we’ll leave
the issue of measuring devices necessarily vague, returning to it
with a vengeance in later sections. Now let’s get back to a state-
ment of the Measurement Problem itself, which is composed of
the two following commonsense queries:

QUANTUM MEASUREMENT PROBLEM

A. At exactly what point in the measurement of the electron’s spin
does the probability set “collapse”?

HOW REAL IS THE ''REAL WORLD''?

433

B. How does the act of observing the electron’s spin collapse the set
of likelihoods f

To see just how strange and puzzling, not to mention philosoph-
ically troubling, this Measurement Problem really is, let’s take a
moment to discuss these points in somewhat more detail.

In everyday life when we think of making a measurement —
say, measuring the size of our living room for a new carpet — the
moment at which the measurement occurs seems self-evident. Or
does it? For example, does the measurement occur at the precise
instant when we lay the yardstick down for the last time on the
other side of the room? Or does it occur when the result of the
measurement impinges on our consciousness? Or did it occur
before we ever laid the ruler on the floor, perhaps when we first
decided to make the measurement? Common sense would proba-
bly argue for the first alternative, but if there’s one thing that
physicists have learned about the world of the quantum, it’s not
to trust everyday, macroworld common sense. And when we de-
scend to the level of quantum objects, the situation doesn’t get
any easier. For instance, at a large experimental particle-physics
laboratory like CERN in Geneva, a particular experiment de-
signed to measure an attribute of a quantum object may go on
for months. So even here we are faced with the problem of ex-
actly when the measurement of the attribute takes place. Is it
when the experiment is planned? When the accelerator is turned
on? When the ghostly tracks of the particle are seen in a bubble
chamber? The fact is, no one really knows. And until the situa-
tion can be resolved, the question of when a quantum object ac-
tually acquires its attributes will remain open. And with it will
remain open the kind of reality that underlies the surface world
of observed phenomena.

Equally troublesome is the second point, the mechanism by
which a physical measuring device acts to collapse a metaphysi-
cal wave of probabilities. For the sake of concreteness, let’s as-
sume that an ordinary meter stick qualifies as a valid
“collapsing device.” How could it be that such a material device,
when used to measure the position of an electron (admittedly
this is a very finely graduated meter stick), could act upon the
quantum wave function, an object composed of pure information
with no tangible material reality at all? Or, put the other way
around, how could such an ephemeral object as a wave of proba-
bility (i.e., information) act to give tangible physical attributes

434

PARADIGMS LOST

like position or spin to material objects? In the sections that fol-
low we’ll explore a number of competing answers that have been
offered by the quantum theory community. But at the moment
we have other fish to fry.

Since it bears significantly upon the Quantum Measurement
Problem, at this point I’d be remiss if I didn’t make at least a
small gesture of obeisance in the direction of the famous Heisen-
berg Uncertainty Principle. No account of quantum phenomena,
popular or otherwise, can omit this most striking of results, if
for no other reason than its enthusiastic application, as well as
concomitant misunderstanding, within a wide range of disci-
plines from modern physics to modern art and a lot in between.

Recall from Schrodinger’s solution to the Description Problem
that for every attribute A there is a family of waveforms that
goes along with A , given by the waveform-attribute dictionary.
As with most dictionaries, the converse is also true here: With
every family of waveforms, there is a corresponding quantum
attribute. Let’s call this fact the Dictionary Correspondence
Theorem. The “attribute” might not be anything to which we
would ordinarily be able to attach physical significance or mean-
ing; nevertheless, in the abstract space of attributes it has full
voting rights along with more familiar citizens like position, mo-
mentum, and all the other celebrity attributes. This situation is
graphically illustrated in Figure 7.5 by the family of “piano
waves,” which correspond to an as yet unnamed attribute. Per-
haps in honor of my amateur pianist neighbor, we could dub this
attribute “out of tune.” Anyway, the point is that there is a
duality between attributes and waveform families. Just as the
Supreme Court has decreed “one man, one vote,” in the quan-
tum world the laws of Nature are equally strict and dictate that
to every attribute there is a waveform family and conversely.

Now suppose we are given a particular waveform family
[wi(x, 01, » = 1, 2, . . . , together with its associated attribute
A . Mathematical fact tells us that there is something else besides
A that we can associate with the given waveform family: an-
other waveform family that is as unlike the family \w,(x , 0} as
a family can be. Let’s call this the waveform family conjugate to
[wi(x, 01, and denote it by j m^x, 01, t = 1, 2, . . . But by the
Dictionary Correspondence Theorem, associated with this conju-
gate waveform family jm.O, 01 is an attribute V, which is in

HOW REAL IS THE ''REAL WORLD''?

435

some sense as “unlike” the attribute A as possible. There is an
important mathematical relationship between the two waveform
families associated with the conjugate attributes A and V. To
describe this relationship, it’s necessary to recall our earlier met-
aphor about prisms and waveforms.

Suppose we have two prisms, one for the attribute associated
with the waveforms W and the second for the attribute paired
with the family M (here and in what follows, I have abbreviated
the family names to bold symbols for ease of writing). Now let’s
pass an arbitrary waveform family X through the W prism. We
will obtain a “W rainbow” consisting of Ny/ colors. The number
Nw is an inverse measure of how closely the family X resembles
the prism family W; i.e., if A’w is large, the resemblance is small,
and conversely. Similarly, if we pass the waveform family X
through the M prism, we obtain an “M rainbow” composed of
JM colors inversely measuring the resemblance of the X family
to the M family. The crucial mathematical fact about this situa-
tion is that the product X is always greater than zero.
In fact, it can be shown that there is a constant R > 0, such
that the product NwiVM > R. And this constant R is indepen-
dent of the particular waveforms. As a technical aside, the spe-
cific value of R depends upon the particular units used in the
problem and is not too important for us. What is important is
that R is always fixed by those units and is always bounded
away from zero. In the jargon of mathematics, the foregoing
relationship between Nw and Nyi is termed the Spectral Area
Theorem, and in plain language it merely states the fact that two
prisms corresponding to conjugate waveforms (hence, to conju-
gate attributes) cannot each resolve the same waveform family
X to an arbitrarily fine degree of precision. There is some irre-
ducible level of coarseness in the joint resolution of the family
X, with the joint uncertainty in the overall resolution being
bounded from below by R. The celebrated Heisenberg Uncer-
tainty Principle is a direct consequence of this Spectral Area
Theorem, which holds for any two prisms corresponding to con-
jugate waveform families W and M and any third family X.
Let’s see why.

The commonly held view of the Heisenberg Uncertainty Prin-
ciple is that it involves an irreducible disturbance, or uncer-
tainty, introduced into the measurement of one attribute due to

436

PARADIGMS LOST

the intrusion of the measuring device when making a measure-
ment on a different attribute. To illustrate this misleading idea,
suppose we have a ball rolling along a straight line and we want
to measure its current position. One way to do this would be to
take a fast-frame photo of the ball, thereby “freezing” its posi-
tion at some instant. But to do this we would have to bounce
photons off the ball in order to get the picture, and the photons
would necessarily impart some energy to the ball, thereby dis-
turbing its velocity at the moment in question. We might argue
that the influence of a photon or two would be insignificant,
which is true enough — for ordinary footballs or baseballs. But if
the “ball” is an electron or another quantum object, the photon’s
impact does wild and woolly things to the ball’s speed and direc-
tion of motion. The upshot of this entire chain of reasoning is
that the more accurately we want to measure the ball’s position,
the more uncertainty we have to be willing to accept in our mea-
surement of its velocity. This, in a nutshell, is the distilled es-
sence of the popular view of Heisenberg uncertainty: We can’t
simultaneously measure two conjugate attributes with perfect
accuracy, nor can they both have well-defined values at the same
moment.

Probably on account of this picturesque, easy-to-understand
idea that measurement seems necessarily to involve physically
intruding upon the object being measured, the idea grew up in
the world outside physics that the cause of Heisenberg uncer-
tainty can be laid at the doorstep of the measurement act itself.
To illustrate the popular view, allow me to quote from a recent
popular book purporting to describe the Uncertainty Principle:

In the subatomic world, the act of measurement changes the sys-
tem being measured, giving rise to what is known as the Heisen-
berg Uncertainty Principle. The principle tells us that if we
choose to measure one quantity (e.g., the position of an electron),
we inevitably alter the system itself and therefore can’t be certain
about other quantities (e.g., how fast the electron is moving).

This interpretation is just plain wrong. Leaving aside the fact
that not every measurement act involves a physical interaction
with the system, the common misconception described here holds
only when the attributes involved are what we have termed con-
jugate. So, for example, there’s no particular problem (at least
in principle) in arbitrarily accurate simultaneous measurements

HOW REAL IS THE ''REAL WORLD''?

437

of both a particle’s position and its energy, as position and en-
ergy are not conjugate attributes. Since it’s now evident that the
physical act of measurement in and of itself has nothing to do
with causing the measurement uncertainty noted by Heisenberg,
what is the basis of this striking principle of ignorance? From
what has gone before, the proverbial perceptive reader will by
now be sensitized to the claim that the rock-bottom cause lies in
the Spectral Area Theorem. Let me sketch the argument.

Let’s suppose we want to measure some attribute A , like posi-
tion. Using our prism metaphor, we know that the attribute A
has its own special prism. From our earlier discussion, we also
know there is a waveform family A associated with the attribute
A . In addition, we automatically obtain free of charge a conju-
gate attribute V, with its own special prism and its own wave-
form family V. The Spectral Area Theorem tells us that if X is
any waveform family whatsoever, corresponding to its own at-
tribute X, if we pass the family X through the A and V prisms,
the “rainbows” emerging must satisfy a relationship that, in ef-
fect, says that if there are a lot of colors in one of the rainbows,
then there can only be a small number of colors in the other, and
conversely. Here it’s crucial to note that the number of colors
that come out of a prism is an inverse measure of how good a job
that prism does in pinning down (i.e., measuring) the values of
the attribute X. But since this inverse relationship must hold
for any waveform family X, i.e., for any attribute X, let’s just
take X = A. In this case, by passing the waveform family for A
through its own prism we will naturally get a rainbow with only
a small number of colors, since that’s what the A prism is de-
signed to do when faced with the waveform family A. But the
Spectral Area Theorem now requires that passing the conjugate
waveform V through the A prism will give a rainbow with the
maximal number of colors; i.e., the A prism will not be able to
pin down values of the attribute V at all! Clearly the argument
is the same if we interchange the roles of A and V, taking X =
V instead. The dilemma is that we have only one prism with
which to do our resolving, and that prism is terrific at resolving
only its design-type of waveform and awful at resolving the
waveform conjugate to it. This is the real meaning of the Hei-
senberg Uncertainty Principle, and it should now be clear why
there is at least no theoretical obstacle to simultaneous perfect
measurements of two attributes that are not conjugate. Since the

438

PARADIGMS LOST

Spectral Area Theorem applies only to conjugate attributes, if
the attributes in question are not conjugate there is no Spectral
Area Theorem, hence no Heisenberg uncertainty. Having paid
our respects to the genius of Heisenberg, let’s now move on to the
other mountaintop — the Quantum Interpretation Problem.

We found the top of the first quantum mountain littered with
all the problems of measurement just considered. These issues
all center upon the meaning of observation, and what precisely
an act of measurement can do in the way of generating knowl-
edge about the dynamic attributes of a quantum object. By way
of contrast, at the top of the second quantum mountain lie scat-
tered a plethora of problems concerning the properties of a
quantum object when it’s not being measured. In short, what we
find is the question: To what degree does a quantum object pos-
sess any dynamic attributes when it’s cavorting about in its
birthday suit, blissfully unobserved? Both quantum theory and
experimental quantum fact support the position that a quantum
object like an electron behaves like a wave when it’s not being
measured, and that it behaves like a particle when a measure-
ment is made. Thus we can state the main question as:

THE QUANTUM INTERPRETATION PROBLEM
What is the true “nature” of an unmeasured quantum object f

The best way to see what we mean here by the term nature is to
examine what might be called the orthodox and the reactionary
schools of thought on the Interpretation Problem.

Orthodox View

1. The wave function gives a complete description of any single
quantum object.

2. All quantum objects represented by the same wave function
are physically identical.

3. The information an observer lacks about an unmeasured ob-
ject is simply not there to be known.

4. The observed differences between identical unmeasured objects
are due to inherent, i.e., quantum, randomness in the objects.

Reactionary View

1. The wave function gives only a statistical description of an
ensemble of quantum objects, hence a necessarily incomplete
description of any single such object.

HOW REAL IS THE ''REAL WORLD''?

439

2. Quantum objects represented by the same wave function may
not be physically identical.

3. The observer’s ignorance about the attributes of an un-
measured object is due to the effect of certain “hidden” varia-
bles, which quantum theory conceals from view.

4. Objects with the same wave function may show differences
upon observation because they were physically different
before the measurement.

Those swearing an oath of fealty to the reactionary creed are
often called hidden variables theorists for the obvious reason that
they cling to a classical view of reality. Their credo is that once
the properties and values of these hidden variables are known,
then all the uncertainty about the values of attributes will fade
away, and the quantum object will be seen as no different from a
Newtonian particle. The primary motivation for this vision of
reality is the desire somehow to avoid placing the measurement
process upon a pedestal of special honor among the myriad phys-
ical actions that the universe might allow. The key assumption
separating these two views of reality is the second point on each
list: the contention that there is a one-to-one correspondence be-
tween the grass-roots physical reality of dynamic attributes for
objects, and the hard-to-get-your-hands-on mathematical reality
of wave functions.

Before going into the courtroom, I think it’s worth noting
that the vast majority of working physicists are neither ortho-
dox nor reactionary, but pragmatic. The typical physicist in the
lab just is not bothered by these ontological questions, and re-
gards quantum theory solely as a “machine” for making predic-
tions about the world of phenomena. Thus the mountaintops of
the Measurement and Interpretation Problems hold no fascina-
tion for him, since they are concerned with the Shangri-La of
deep reality, not with the dusty flats of observed phenomena. As
long as he can use the quantum machine to describe and predict
the results of his experiments, the average physicist is just like
the average car owner: He doesn’t care what makes the magic
work. He just wants to know what levers to pull and what knobs
to twist in order to get from A to B. Fruitful as that attitude is
in the world of phenomena, it takes us no closer to an under-
standing of what kinds of miracles underlie the workings of the
machine. Ultimately it’s at this level that the battle must be
fought, and the strategies employed are completely dependent

440

PARADIGMS LOST

upon the attitudes the expedition leaders take toward scaling
our twin peaks, Mount Measurement and Mount Interpretation.
Before turning the floor over to the various climbers and an ac-
count of their strategies for reaching the summits, let’s briefly
review the impressive volume of vocabulary introduced in the
preceding sections. The main items are compactly summarized in
the box below.

TERMS AND CONCEPTS

quantum object an object of any size that displays both
wave and particle behavior in the quantum manner
static attribute a property of a quantum object that doesn’t
change over time, such as mass, charge, and spin
dynamic attribute a time-varying property of a quantum
object, like position, velocity, energy, and spin axis
orientation

wave function a mathematical object displaying wave
behavior that encapsulates all the attributes of a
quantum object

waveform family a collection (usually infinite) of waveforms
sharing certain characteristics that enable them all to be
associated with one dynamic attribute

Measurement Problem the question of how and when the act
of measurement “collapses” the wave function
Interpretation Problem determination of the nature of a
quantum object when it is in its unmeasured state
Uncertainty Principle Heisenberg’s assertion that conjugate
attributes cannot both be simultaneously measured to an
arbitrary degree of precision

hidden variables postulated variables hidden from
observation whose values, if known, would account for
measurement uncertainty

With this lexicon at our side, let’s give the podium over to the
Prosecution and its parade of witnesses claiming that when it
comes to objective reality independent of an observer, there just
isn’t any such thing.

HOW REAL IS THE ''REAL WORLD''?

441

THE ROMANTIC REALITIES

Not far from my old apartment in the center of Vienna, there’s
one of those raucous, sawdust-on-the-floor, student-hangout type
of beer halls advertising 101 or so brands of brew from around
the world — all of them wet! On odd occasions, fortunately rare,
visitors from North America are intrigued by this place and
want to drop in and sample the waters. So despite my best ef-
forts to dissuade them, the dictates of good hospitality demand
my entry through the portals of this smoke-filled den, at least
for a quick pint or two. To make the best of a bad situation, on
these occasions I try to throw my beer-swilling schilling in a
productive direction by ordering the Danish poison Carlsberg,
telling myself that by doing so I’m at least casting a small vote
for science. Why science? you ask. Well, unlike the competition,
which tends to spend its promotional budget on the sponsorship
of pro football telecasts, auto racing, or some other macho type
of activity, Carlsberg invests in quantum theory! More pre-
cisely, Carlsberg invested in Niels Bohr, the spiritual father of
all quantum theorists, and Niels Bohr used that Carlsberg lar-
gesse (the gift of a rather elegant mansion, no less) to house an
institute for theoretical physics in Copenhagen that served for
decades as the mecca for all quantum theorists. The output from
Bohr’s institute still serves as orthodoxy in the community of
physicists when it comes to the Measurement and Interpretation
Problems, so it’s fitting that we start our account of what I’ve
labeled the romantic realities with a consideration of what’s now
usually known as the Copenhagen Interpretation.

Before outlining the case from Copenhagen, let me set the ter-
minological stage. All of the Prosecution’s witnesses will be pre-
senting realities that are “romantic” in the sense that they come
straight from the fantasy novelist’s pen — literally incredible.
It’s the romantic realities that you’re reading about when you
scan those Sunday supplement accounts of quantum theory as a
basis for mysticism, telepathy, parallel worlds, the dialectic, al-
tered states of consciousness, astral projection, meditation, pyr-
amid power, tarot reading, and all the other subdivisions of the
occult found at your favorite bookshop. With the imprimatur of
such intellectual giants as Bohr, von Neumann, Wigner, Heisen-

442

PARADIGMS LOST

berg, and Schrodinger, who could blame the occultists for ap-
propriating at least the form, if not the content, of the ro-
mantics’ far-out views of what’s really what? Here I’ll try
to remain within the confines of the Measurement and In-
terpretation Problems as outlined above, but if the reader no-
tices the narrative occasionally slipping off the track in the
direction of the occult, it’s only because the romantic reali-
ties suggested by the quantum facts are truly so strange that it’s
sometimes difficult to separate serious science from both
hopeful and hopeless speculation. With these disclaimers on
the record, it’s on to the Little Mermaid and the Tivoli Gardens
of Copenhagen for the testimony of our first romantic, Niels
Bohr himself.

THE COPENHAGEN INTERPRETATION
There is no deep reality

Bohr’s position on reality is simple: There is no deep reality.
Just that. No deep reality of any kind whatsoever. The implica-
tion of such a claim is that quantum objects in their unmeasured
state literally have no dynamic attributes. In contrast to the
pragmatists, who might say that the question of the existence of
such attributes is literally meaningless, the Copenhagen Inter-
pretation developed by Bohr goes much further. Copenhagenists
say that such attributes definitely do not exist. Or, more accu-
rately, whatever attributes objects might possess are contextual:
They depend upon the measurement situation, so they cannot be
ascribed to the object independent of the measuring device and
the act of measurement. This claim gives rise to Bohr’s famous
Complementarity Principle, which states that whether the object
displays wave properties or particle properties depends upon the
measurement situation and not just on the object itself. In other
words, the Heisenberg Uncertainty Principle is an intrinsic
property of Nature, and that the observer, the measuring device,
and the system to be measured form a whole that cannot be di-
vided. More prosaically, we might express this wave-particle
complementarity idea using Bohr’s own phrase: “The opposite
of a big truth is also a big truth.”

Then where do these attributes come from if they don’t exist
for unmeasured objects? Well, if they’re there in the object’s
measured state, then the only place they can come from, accord-

HOW REAL IS THE ''REAL WORLD''?

443

ing to Copenhagenists, is out of the measurement act itself. In
other words, for a Copenhagenist the dynamic attributes are not
a property of either the quantum object or the measuring device
taken separately, but are a property of the joint relationship be-
tween the object and the device. Somehow measurement seems a
little like nitroglycerine: Neither nitric acid nor glycerine is ex-
plosive on its own, but when you bring them together, BANG!
This summarizes the Copenhagen view of attributes, too. Bring
an object together with a measuring device and BANG: instant
attributes.

There are several drawbacks to the Copenhagen view, not the
least of which is that it assigns a privileged role to the measur-
ing instrument. As far as the Measurement Problem is con-
cerned, the Copenhagenists put all the mysteries of the wave
function collapse right at the boundary between the quantum ob-
ject and the measuring device. This leads to the puzzling situa-
tion in which two radically different types of systems are forced
to interact: a classical measuring device and a quantum object. So
in actuality the Copenhagen view doesn’t solve the Measurement
Problem at all, but merely sweeps it under the rug into the one
place that’s inaccessible to all observers — the inside of the mea-
suring device itself. As to the Interpretation Problem, the Co-
penhagenists are clear: An unmeasured quantum object has no
attributes; ergo, there is no deep reality underlying the world of
phenomena. In David Mermin’s words, the traditional Copenha-
gen view answers Einstein by saying, “The Moon really isn’t
there if you don’t look.” It should be noted that more recent
detailing of the Copenhagen view by W. Zurek and others soft-
ens this conclusion somewhat to maybe the Moon really isn’t
there.

Oddly enough, despite the major drawbacks to the Copenha-
gen position, to this day it constitutes the conventional wisdom
of the physics community. One of the reasons is undoubtedly
Bohr’s immense prestige, as well as the fact that his institute
got its oar in the water first in the reality generation game. But
an equally substantive reason is a hard mathematical fact
proved by von Neumann that tends to give support to the Copen-
hagen view. Interestingly enough, although the Copenhagenists
latched on to his result as evidence for their case, von Neumann
himself leaned toward our next romantic reality, the school of
consciousness.

444

PARADIGMS LOST

CONSCIOUSNESS-CREATED REALITY

The observer's consciousness creates reality

As a reaction to the classical/quantum schizophrenia of the
Copenhagen view, von Neumann argued that both the measuring
device and the quantum object should be treated as quantum
systems. Pursuing this symmetry, von Neumann produced an
elegant mathematical basis for quantum phenomena in his 1932
treatise Die Mathematische Grundlagen der Quantenmechanik. In
this magisterial work von Neumann showed that if the predic-
tions of quantum mechanics are correct, then the world cannot
be made out of ordinary objects possessing innate attributes. In
fact, by this result the world cannot even be constructed out of
combinations of unobservable ordinary objects. This conclusion
seems to banish forever any kind of hidden variable theory from
the reality game. The nature of this banishment will be consid-
ered in far greater detail below. As noted a moment ago, this
hard fact was pounced upon by the Copenhagenists as providing
mathematical ammunition for their claims. But the world of
quantum theorists is just as tricky as the world of quantum the-
ory, so things were not nearly so clear-cut as Copenhagenists
might have hoped.

Recall that the Copenhagen position maintained that there
was a definite separation between the measuring device and the
quantum object being measured, and that the wave function col-
lapse was assumed to occur in some vague neighborhood between
the two. Yon Neumann wanted to pin down the size of this
neighborhood. To everyone’s surprise and consternation, when
he put the object and the device on the same footing by thinking
of them both as quantum objects (under rather idealized circum-
stances), von Neumann discovered that as far as the final ob-
served results were concerned, he could put the “cut” between
the two anywhere he pleased! From the standpoint of the Mea-
surement Problem, this means that the wave function collapse
can occur in the system, in the device, or anywhere in between —
take your pick. In the example of measuring our living room
carpet, this would mean that as far as the final quantum-theore-
tic description of our living room goes, we could think of the
measurement as taking place at the moment we decided to buy
the carpet and knew that such a measurement would have to be

HOW REAL IS THE ''REAL WORLD''?

445

made, or at the very moment the actual measurement entered
our minds, or anywhere in between.

As a consequence of this truly shocking result, von Neumann
focused upon the one even slightly questionable (from a rigorous
scientist’s viewpoint) link in the entire measurement chain: the
human mind. Although he never actually said so in print, one
can infer from his many parables and remarks on the matter
that his “Cut Theorem” forced von Neumann into taking refuge
in human consciousness as the final “collapsor” of the wave
function. In this last refuge of quantum theorists, von Neumann
is joined by his fellow Central Europeans Eugene Wigner and
Erwin Schrodinger, who between them cooked up what are prob-
ably the most celebrated and colorful thought experiments in the
annals of the quantum theory — Schrodinger’s Cat and Wigner’s
Friend — to illustrate graphically the difficulties involved. The
point of Schrodinger’s experiment is to illustrate the profound
weirdness of the wave function as a complete description of a
macroscopic object like a cat. The setup for the two thought ex-
periments is shown in Figure 7.6.

The experiment involves a sealed and insulated box (A ) con-
taining a radioactive source ( B ). The source has a 50-50 chance
of triggering the Geiger counter ( C ) during the course of the
experiment, thereby activating a mechanism ( D ) that causes a
hammer to smash a flask of prussic acid ( E ), thereby killing the
cat ( F ). An observer (G) has to open the box in order to collapse
the wave function into one of the two possible states (cat =
DEAD, cat = ALIVE). A second observer (Wigner’s Friend)
(H) is then needed to collapse the wave function of the larger
system comprising the first observer ( G ), the cat ( F ), and the
equipment (A-E). The problem here is that now the original ob-
server (G), Wigner’s Friend (H), and the apparatus (A-E),
plus the cat, constitute a new system, which may itself require
an “Acquaintance” to collapse its wave function, and so on.

Wigner’s interpretation of the foregoing experiment is that
quantum theory breaks down when the conscious awareness of
the observer is involved. For Wigner his own conscious mind is
the basic reality, and the things in the world “out there” are not
much more than useful constructions built out of his own past
experiences, somehow coded into his consciousness. In this pic-
ture of reality, the moment when the information about an ob-
servation enters the consciousness of an observer is when the

FIGURE 7.6. The Schrodinger’s Cat and Wigner’s Friend Experiment

mathematical wave function collapses into physical reality. De-
spite the stature of its supporters, the feeling of most physicists
today when they hear this kind of explanation is aptly summed
up by Stephen Hawking’s remark: “When I hear of
Schrodinger’s Cat, I reach for my gun.” On such an unambigu-
ous note of rejection of consciousness-generated reality from
today’s premier cosmologist, let’s hop across the Atlantic from
the old world of Copenhagen, Budapest, and Vienna to the hill
country of central Texas for our next romantic contender.

THE AUSTIN INTERPRETATION
Reality is observer-created

Texas may call itself the Lone Star State but Texans have al-
ways done things in a big way, so when the agenda item is real-
ity generation no one will be surprised to find that the “lone
star” is magically transformed into an entire universe of glow-
ing objects, the centerpiece being nothing less than the meaning
of meaning itself. The chief architect of this Texas-sized version

HOW REAL IS THE ''REAL WORLD''?

447

of reality is John A. Wheeler, director of the Center for Theo-
retical Physics at the University of Texas at Austin.

The heart of the Austin Interpretation championed by
Wheeler is the idea of a reality created by the observer through
exercise of the measurement option. The Austin school believes
that we are wrong to think of the past as having a definite exis-
tence “out there.” The past exists only insofar as it is present in
the records we have today. And the very nature of those records
is dictated by the measurement choices we exercised in generat-
ing them. Thus, if we chose to measure an electron’s position
yesterday in the lab and recorded the resulting observation, then
that electron’s position from yesterday exists but its velocity
doesn’t. Why not? Simply because we chose to measure the posi-
tion and not the velocity.

Because this very act of choosing is always involved in what
we measure, Wheeler feels the act of observation is “an elemen-
tary act of creation.” In actuality, the Austin Interpretation
doesn’t go quite so far as to claim that these choices dictate the
reality of macro world objects like tennis balls, but rather con-
fines its claims to the microworld of quantum objects like elec-
trons. Nevertheless, Wheeler’s message is clear: “No elementary
phenomenon is a phenomenon until it is an observed phenome-
non.” To illustrate the point, Wheeler has introduced an impor-
tant variation on the classic double-slit experiment discussed
earlier. Recall that in the standard situation we first decide
which of the slits is to be open, then we turn on the projector
and observe the pattern of response at the detectors. In
Wheeler’s Delayed-Choice Experiment, we wait until after the
quantum objects have passed the slits before we decide which
gaps are to be open.

To illustrate the idea, consider receipt of light on Earth from
a distant point source (a quasar) as shown in Figure 7.7. One of
the great theoretical predictions of relativity theory was that the
gravitational field around massive bodies would act to bend
beams of photons as they passed nearby. This is the so-called
gravity lens effect, and it works in much the same way that a
magnifying glass bends light rays here on Earth. Just by
chance, there happens to be a large galaxy standing directly be-
tween Earth and the quasar QSO 0957 + 561. This means that
light from the quasar has to pass around the galaxy in order to
be collected in Earth-based telescopes. In their observations of

448

PARADIGMS LOST

the quasar, astronomers have detected a double image that they
attribute to this bending of the quasar’s light as it passes
around the rim of the galaxy on either side, as shown in the
figure.

From the perspective of delayed choice, the interesting ques-
tion becomes: When we detect a photon from the quasar today
on Earth, which side of the galaxy did it pass around on its trip

HOW REAL IS THE ''REAL WORLD''?

449

here? Common sense would argue that this question must have
been settled billions of years ago when the photon moved past
the galaxy, since the quasar is so old that the light started on its
way to us even before our sun began to shine. But remember the
rule: Never trust earthly common sense when it comes to quan-
tum objects. The Austin crowd says no, we can actually influence
what we have a right to say about the past by what we choose to
measure today. Here’s how. First, simply use standard optical
means to bring together the two beams that have gone around
the two sides of the galaxy. Then allow them to cross. Now exer-
cise the measurement option by deciding whether to put your
detector at the intersection point A, or at B where the beams
have again separated. This option can be exercised differently
for each photon, but only one choice per photon, please! If you
choose the first option, interference effects indicating the photon
took both paths will be seen; the second choice will show that the
photon took only one of the two paths around the galaxy. Thus
the Delayed-Choice Experiment seems to show that which path a
photon took around the galaxy billions of years ago is dictated
by the measurement choice we make here on Earth today. In this
way, Wheeler argues, the observer creates reality.

We should hasten to note that the Austin Interpretation
champions an observer- created reality, not a consciousness-cre-
ated one. The Austin view, while differing from Copenhagen in
significant ways, still accepts some of the crucial aspects of
Bohr’s position. Most important, the two schools agree that sci-
entists can communicate unambiguously only about the final re-
sults of a measurement. For Wheeler, the essence of existence
(reality) is meaning, and the essence of meaning is communica-
tion defined as the joint product of all the evidence available to
those who communicate. In this view meaning rests on action,
which means decisions, which in turn force the choice between
complementary questions and the distinguishing of answers.
Putting all these links together, out pops the Austin Interpreta-
tion of reality generation by exercise of the quantum measure-
ment option.

Of course the reliance upon an observer to create reality is
also a part of the von Neumann-Wigner consciousness school of
romantic realities. However, the Texans are very clear on the
point that their brand of reality has no need to invoke the spe-
cial role of consciousness. They endorse the Copenhagen view

450

PARADIGMS LOST

that what constitutes a measuring apparatus is any device that
records a quantum phenomenon, giving rise to Wheeler’s state-
ment “Let’s not invoke consciousness as a prerequisite for what
in quantum mechanics we call the elementary act of observa-
tion.”

To summarize the Austin position on the twin problems of
measurement and interpretation, the stance is clear on the mat-
ter of the nature of unmeasured quantum objects: These objects
have no attributes until a measurement is taken; i.e., there is no
objective reality without measurement. As for the Measurement
Problem, the Austin Interpretation seems to be pretty much in
agreement with that of the Copenhagenists: Possibility becomes
actuality at the moment the record is made. In attempting to pin
down when this moment takes place, the Austin Interpretation
invokes the communication postulate, which seems to imply that
the wave function collapse occurs when the elementary quantum
process is brought to a close by an irreversible act of amplifica-
tion. This act of communication closes Wheeler’s Meaning Cir-
cuit of existence, shown in Figure 7.8. Here the quantum aspects
of existence appear in the “underground” part of the loop back
from meaning to physics. Wheeler’s self-referential universe
logo, also shown, neatly sums up the Austin view of the universe
as a Delayed-Choice Experiment in which the existence of ob-
servers who see what’s happening gives tangible reality to every-
thing else.

By now it should be evident that problems of language are
becoming increasingly severe as we try to bring the testimony of
the Prosecution’s romantic realists into some semblance of con-
tact with ordinary images of space, time, matter, and all the
rest. One of the first to try to come to grips with this language/
reality gap is our next witness, Werner Heisenberg, who cham-
pions a reality in which what’s real is a combination of what
may be and what is. Let’s swear him in.

THE DUPLEX INTERPRETATION

Reality consists of potential and actuality

We’ve repeatedly emphasized that the quantum wave function
somehow encapsulates all the possible attributes a quantum
object can display — once we finally get around to making a mea-
surement. After years of pondering Feynman’s forbidden ques-

HOW REAL IS THE ''REAL WORLD''? 451

tion, Heisenberg eventually concluded that reality consists of
two disjoint worlds: the world of potential ( potentia ) and the
world of actuality, with the two joined by the act of measure-
ment. How does his vision differ from that put in the record by
the previous distinguished witnesses for the Prosecution, who
themselves also sanctified the measurement act?

For Heisenberg the only reality, as that term is usually em-
ployed in ordinary life, is the world of actuality, i.e., the world
of phenomena. But phenomena have to be constructed out of
something, don’t they? What is that something? In the Duplex
Interpretation that something underlying tangible reality is

452

PARADIGMS LOST

pure potential, the tendency for things to come out one way and
not another once they are observed phenomena. Thus Heisenberg
is here taking the wave function at face value, saying that this
will-’o-the-wisp realm of potential comprises the “stuff” out of
which things like knives, forks, plates, tables, chairs, and me-
dium-rare steaks are ultimately formed. So the unmeasured
world literally is just what the quantum wave function repre-
sents it to be — a world of unrealized potential. At the moment of
measurement, one of these tendencies is magically granted a
more exciting life-style, being transformed into the world of ac-
tuality as an observed phenomena. It is at this moment that
whatever attributes were implicit in the potentia surface as real
attributes.

At first glance Heisenberg’s universe of potential looks a lot
like the world of potential in Las Vegas, where the roulette
wheel has the potential of displaying any of thirty-seven out-
comes before the croupier decides to give it a spin. But, in fact,
Heisenberg’s potentials are much less well defined than this. For
him even the range of possibilities is not set until you specify
the measurement option. Referring again to the casino setting,
the potentia would be represented by the entire world of possibil-
ities for all the different games and devices the casino offers, like
craps, twenty-one, roulette, and keno. The possibilities present
in a particular wave function would become specified only once
we decide which game we’re going to play, i.e., once we’ve de-
cided upon the specific measurement situation.

To emphasize the “unreality” of quantum objects in the un-
derworld of potentia, Heisenberg states:

In the experiments about atomic events we have to do with things
and facts, with phenomena that are just as real as any phenomena
in daily life. But the atoms or the elementary particles themselves
are not as real; they form a world of potentialities or possibilities
rather than one of things or facts. . . . Atoms are not things.

Like the other romantics, Heisenberg disavows any sort of ob-
jective, observer-independent reality propping up the world of
everyday phenomena. The world of potentia cannot really be seen
as anything other than a kind of shimmering mirage of dream-
like reality, waiting to be awakened into actuality by the magical
Midas touch of measurement. But let’s consider for a moment
some of the differences between potential and actuality as seen

HOW REAL IS THE ''REAL WORLD''?

453

by Heisenberg, and the realities put forth by his Prosecution
cohorts.

First of all, the Measurement Problem. Given his association
with Bohr and the Copenhagen school, it would be unthinkable
to find Heisenberg’s position on wave function collapse depart-
ing from the orthodox Copenhagen line. Indeed, the Duplex
Interpretation, just like the Copenhagen, Austin, and Con-
sciousness interpretations, asserts that there’s something very
special indeed about making measurements. Furthermore, all in-
dications are that Heisenberg was just as vague as the Copen-
hagenists as to exactly when this sacred event takes place.
However, his position does insist upon the exercise of the mea-
surement option before the possibilities inherent in the wave
function can actually be specified, a stance that puts the Duplex
and Austin positions into momentary conjunction, an alliance
that is soon broken by the Austin insistence on meaning as the
core of all existence. The Duplex Interpretation seems to have
nothing at all to say on this key point, a fact that also separates
it from the advocates of consciousness-created realities.

When it comes to the Interpretation Problem, the Duplex po-
sition is unambiguous: Quantum objects have no meaningful ex-
istence other than in the world of potentia, and they certainly
don’t possess anything that could be called attributes in the un-
measured state. So again we are faced with testimony arguing
that objective reality is a physical fiction brought on by our lack
of linguistic sophistication and inability to comprehend what it
could possibly be like to live in a world of pure potential. But
like all the other romantics heard from so far, Heisenberg pro-
poses a world in which there are two halves separated by the
all-embracing and all-consuming act of measurement. It’s the
task of our last Prosecution witness to convince you that per-
haps the Measurement Problem is no problem after all — pro-
vided you’re ready to entertain the idea of lots of realities
instead of none at all.

THE MANY-WORLDS INTERPRETATION

There is a universe for every possible observation,
each of them equally real

In 1941 the Argentine writer and poet Jorge Luis Borges pub-
lished a small volume of fantastic stories, The Garden of Forking

454

PARADIGMS LOST

Paths. In the title story, the sinologist Stephen Albert tells the
protagonist, Hsi P’eng, about the infinite labyrinth of Hsi’s an-
cestor who, according to Albert,

did not think of time as absolute and uniform. He believed in an
infinite series of times, in a dizzily growing, ever spreading net-
work of diverging, converging and parallel times. This web of
time — the strands of which approach one another, bifurcate, inter-
sect or ignore each other through the centuries — embraces every
possibility. We do not exist in most of them. In some you exist
and not I, while in others I do, and you do not, and in yet others
both of us exist.

This phantasmagorical Borgesian world was brought to the au-
gust pages of Reviews of Modern Physics sixteen years later when
Hugh Everett III, a student of Wheeler’s, published his doc-
toral dissertation, leading to what is now termed the Many-
Worlds Interpretation of quantum theory.

Everett started from the same place as von Neumann in that
he regarded both the system that was to be measured and the
measuring device as quantum objects. But then instead of wor-
rying about when the wave function collapses, Everett said, in
effect, forget about the collapse. Following this dictate to its log-
ical conclusion, Everett’s theory claims that whenever the sys-
tem and measuring device interact, the new system composed of
the two splits into as many copies as there are possible outcomes
of the measurement. So if the measurement could have yielded
one of M possible outcomes, after the interaction Everett says
there are now M equally real “worlds.” In World 1, the measur-
ing device shows possible outcome number one; in World 2, it
shows possible outcome number two; and so on. Consequently,
instead of the wave function’s collapsing, the quantum system
realizes all possible outcomes, and each of them is actually real-
ized in its own separate world. At this juncture, the practical
man poses the obvious question: “If there are so many different
worlds out there, each of them real, why do I seem to see only
one of them (at a time, at least)?”

Everett’s answer to the above query is one that will brighten
the day of every science fiction writer, mystic, and modern cos-
mologist: The inhabitants of these worlds live on parallel planes
of existence. So the sci-fi writers are on the right track after all,
and there is a universe in which the Confederacy did win the

HOW REAL IS THE ''REAL WORLD''?

455

Civil War and another in which UCLA didn’t win even one
NCAA basketball title. And we have the final authority of the
physicists to assure us that these worlds really are out there.
But if you’re a USC man, don’t start making plans yet for emi-
grating to that UCLA-less heaven. There is a censor in the cos-
mos who ensures that we humans can occupy only one universe
at a time. So we can’t see all the others, even though Everett
assures us they’re out there, and each of them is just as real as
the universe we actually experience. Just why we should be con-
fined to a single universe at a time is anybody’s guess, and most
commentators on the matter (non-science fiction writers, that is)
fall back upon that most Chomskian of explanations that we hu-
mans just happen to be wired up that way.

Popular expositors of the Many-Worlds Interpretation have
obviously fallen in love with the idea of a myriad of universes
continually branching away from each other as each observation
takes place. Somehow the idea of 10100+ universes does capture
the imagination. In all fairness to Everett, however, it should be
noted that he thought of the situation in slightly less romantic
terms, in which only the measuring apparatus itself branches
into these different possibilities. Of course this in itself is a
pretty wild notion, but it pales by comparison with the popular
image of branching universes rather than branching meter
sticks or Geiger counters. An equivalent, but even less romantic,
view is offered by David Deutsch, who thinks of there being a
fixed, but infinite, number of universes at all times. In this
setup, whenever a measurement is made this infinity of uni-
verses just reconfigures itself to account for the possible experi-
mental outcomes. Thus, rather than an ever-increasing temporal
sequence of Borgesian parallel worlds, the Deutsch picture is
more like that suggested much earlier by Boltzmann in which
the worlds all exist simultaneously and always have. They some-
how just occupy different “spaces” that are mutually inaccessi-
ble to each other.

Whichever way you call it, the Many-Worlds Interpretation,
despite its frankly bizarre character, is a favorite with a number
of physicists for several reasons. The most important is that it’s
the only reality that doesn’t sanctify the measurement act. As
far as Everett’s thesis goes, measurement devices and actions
exist on the same footing as any other physically realizable ac-
tivity. Since there is no collapse of the wave function, there is no

456

PARADIGMS LOST

Measurement Problem. Beyond any shadow of a doubt, this is
the cleanest possible solution to the Measurement Problem: Just
banish it from the realm of problems by pulling the rug out
from under the essence of the difficulty — the collapse of the wave
function. But the price we pay for this solution is our willing-
ness to accept a resolution of the Interpretation Problem that
stretches credulity. Instead of asking us to picture unmeasured
quantum objects that have no definite existence (as in all the
other romantic realities), Everett jumps to the other end of the
scale and says not only do they really exist, but there are an
uncountably large number of them. It’s hard to imagine con-
cluding the testimony for the Prosecution on a more flamboyant
note than this.

So we’ve come to the end of the Prosecution’s case, and what a
case it’s been: The opposite of a truth is a truth; reality comes
straight out of consciousness; observers dictate what’s real; real-
ity is potential; anything that can happen does happen. No won-
der journalists love these romantic realists. Just as in literature,
where there are many shades of romanticism, so it is in physics,
with the romantic physicists conjuring up many answers to the
Quantum Measurement and Interpretation Problems. For future
reference, the answers proposed are summarized in Table 7.3.

The Defense is going to have a tough time putting up a battle
against such an armada of dazzling realities and intellectual
muscle. While their visions of reality have somewhat less flair
than those of the Prosecution, and involve a slightly more pedes-
trian view of the cosmos and more legwork in developing the
details, the Defense experts are no slouches themselves when it
comes to playing the reality game. So let’s now give our atten-
tion over to their arguments for why there may be something to
the idea of objective reality after all.

THE DOGWORK REALITIES

In his account of his years as Einstein’s assistant at the Insti-
tute for Advanced Study, Abraham Pais tells of an incident in
1948 when Niels Bohr was visiting the institute. Since he didn’t
like the large office assigned to him, Einstein generally used the
smaller office next door that had originally been allocated to his
assistant. Consequently, during his visit Bohr was using. Ein-

HOW REAL IS THE ''REAL WORLD''?

457

SCHOOL

WAVE FUNCTION
COLLAPSE

UNMEASURED

ATTRIBUTES

Copenhagen

by measuring device

don’t exist

Consciousness

by conscious mind

don’t exist

Austin

from communication

created by meter option

Duplex

from measurement act

only phenomena are real

Many Worlds

there’s no collapse

all possibilities are real

TABLE 7.3. The romantic realists

stein’s large office and pondering once again his decades-long de-
bate with Einstein over what we have termed the Quantum In-
terpretation Problem. As Bohr paced the room muttering in his
inimitable way, the one word that came through clearly was
“Einstein . . . Einstein . . As he stood for a moment peering
out the window with his back to the door, Bohr, once again mum-
bled the magic name, and at that very moment Einstein silently
entered the room from next door to get some tobacco from the
humidor on his desk. After a moment, Bohr turned around and
saw Einstein standing there like a genie conjured up by his
magic incantation. Following a few seconds of astonishment, the
two old adversaries both broke out laughing at this seeming
stroke of synchronicity. In quantum terms, Bohr might have
said that only the intervention of his “measurement device”
brought Einstein into the room’s reality. Einstein, no doubt,
would have claimed that he existed all along as a “hidden varia-
ble,” and that once his “value” was known by Bohr’s observa-
tion, all the mystery of the situation disappeared. This little
vignette illustrates the position of the first and by far most
prominent witness for the Defense, Albert Einstein, the univer-
sally acknowledged king of post-Newtonian physics. We’ve al-
ready described Einstein’s position on the matter of reality as
naive realism. For the sake of completeness, let’s briefly summa-
rize its content.

NAIVE REALISM

Deep reality consists of ordinary objects

Since we’ve already considered the realist position in some de-
tail, it suffices now to note that the idea of an ordinary object is

458

PARADIGMS LOST

exactly what the words imply: an object whose attributes really
exist whether or not one is observing them. Naturally, in the
classical view of attributes, the act of measuring a quantum ob-
ject is no more sacred than the act of measuring your living
room for a carpet. It’s just a confirmation of something that
existed all along. Thus for the naive realist the solutions to the
problems of measurement and interpretation are quite clear-cut.

First of all, measurement. Since attributes uniquely exist at
all times, the wave function description is incomplete. This
means there must be hidden variables whose values, when
known, in effect collapse the wave function into a single possibil-
ity. Consequently, the Measurement Problem is the result of the
incompleteness of the quantum description, and disappears as
soon as the additional variables are accounted for. As for the
Interpretation Problem, the Einstein position is equally clear:
All attributes exist at all times, observed or not. So there does
indeed exist a single, objective, observer-independent reality.
End of story.

In the realist picture, attributes like position and velocity
combine in ordinary ways to form new attributes. So, for in-
stance, if two particles collide to form a single new entity, their
respective velocities before the collision can be added to deter-
mine the velocity of the new particle. All operations of this kind
in which attributes are combined involve the Boolean logical op-
erations of AND, OR, NOT and so on. Our next Defense witness
claims that quantum objects may have real attributes after all,
but the logic of the quantum world is just different from that we
normally use.

QUANTUM LOGIC

The quantum world uses a nonstandard type of logic

Shortly after publication of his quantum bible in 1932, von Neu-
mann and Garrett Birkhoff invented a new type of logic that can
be used for describing how quantum objects combine their at-
tributes to form new attributes. To explain the basic idea of
these nondistributive lattices, imagine we have a set of objects
that can possess three sorts of attributes. Call these attribute
types X, Y, and Z. Using normal rules of logic, we can combine
these attribute sets in various ways. For instance, we can form a
new set composed of those objects that possess both attribute X

HOW REAL IS THE ''REAL WORLD''?

459

AND attribute F, called the intersection of X and Y and denoted
Xn F. Similarly, the set of those objects having attribute X OR
F is termed the union of X and F, denoted Xu F. One of the
most important laws of normal logic is the so-called distributive
law, which states that

X OR ( F AND Z) = (X OR F) AND (X OR Z)

X u (Y n Z) = (X u F) n (X u Z)

To illustrate the foregoing relationships, suppose the attrib-
utes in question are the types of polarization that can be dis-
played by a photon. Polarization is an attribute that is
associated with a particular direction in space, and any given
photon is either completely polarized in that direction or com-
pletely polarized at right angles to that direction. Thus the po-
larization attribute can take on only one of two values relative to
a given direction. Suppose three directions are given: vertical,
horizontal, and diagonal, denoted V, H, and D, respectively.
Here, as the names imply, the directions H and V are assumed
to be at right angles to each other, while D is intermediate be-
tween the two. Using the above notation for union and intersec-
tion, we could describe those objects polarized in either the
horizontal or vertical direction as H u V, while those polarized
in both directions would be expressed as II n V (of course, this
set is empty here since photons cannot be completely polarized in
two orthogonal directions at once). The distributive law now
states that the collection of all those photons polarized both hori-
zontally and diagonally plus those polarized vertically consists
of those that are vertically or horizontally polarized plus those
that are vertically -or- 4iagonally polarized. Simple, ordinary
common sense, right? Well, we already know about the value of
common sense in the quantum world.

Let’s consider the Three-Polarizer Paradox as an example of
the failure of the distributive law in the quantum jungle. Recall
that a polarizing filter is just a slab of material that passes only
a single type of polarized light. A good example is a pair of po-
larized sunglasses that pass only visible light in a single direc-
tion, screening out the light from other directions that is the
cause of annoying glare. Suppose we have three such filters, each
designed to pass light polarized in one of the above directions H,
V, and D. The experimental setup is displayed in Figure 7.9.

460

PARADIGMS LOST

When we use only the H and V filters, all light is blocked, re-
flecting our earlier remark that light cannot be polarized in two
directions at once. Now let’s throw a joker into the deck and
consider the third filter, D.

The D filter passes light polarized diagonally to that passed by
either the V or H filter. If we place the D filter either before or
after the H -f V filter stack, the result is just what we’d expect:
No light gets through. However, if we place it in between the H
and V filters as shown in the diagram, a miracle occurs. Light
gets through the stack. How can this possibly be? According to
normal Boolean logic, it can’t. The only “logical” explanation
emerges when we pass to a non-Boolean kind of reasoning in
which the above distributive law is no longer valid. This is the
essence of the argument made by the quantum logicians: Things
are just different in the quantum realm, including the kind of
logic that underlies the combining of attributes of quantum ob-
jects. According to the non-Boolean view, the quantum world
consists of individual islands on which the ordinary rules of
logic apply (the case of individual attributes). But these islands
combine their attributes in a way that can be described only by
some weird, nonstandard rules applicable solely to the world of
the quantum. It’s as if you had a group of Pacific islands on
each of which the natives speak the same language. But when
the islanders get together for their joint annual festival, the
only language allowed is an ancestral tongue in which some of
the ordinary grammatical rules no longer apply. Assume for the

HOW REAL IS THE ''REAL WORLD''?

461

sake of argument that the quantum logic picture is correct.
What does it say about our benchmark tests, the Measurement
and Interpretation Problems?

A major spokesman for the quantum logicians is David Fin-
kelstein of the Georgia Institute of Technology, who argues that
there is nothing strange about the idea of unmeasured quantum
objects’ possessing definite attributes at all times. What is
strange is the way these attributes are combined to form what
we see with our measuring instruments (such as the three pola-
rizers). Thus, on the Interpretation Problem the logicians say
yes, there is an objective reality consisting of quantum objects
having definite attributes. As to the Measurement Problem,
quantum logic says nothing about the when, why, or how of the
wave function collapse, but speaks only about the properties of
the object in the unmeasured state. Thus quantum logic offers us
no help whatsoever in scaling this peak of the quantum terrain.
On this disappointing note, let’s call in our next witness for the
Defense.

While von Neumann’s proof against hidden variable realities
seems to close the books on the naive realist position, dealing
what looks like a mortal blow to the gleam in Einstein’s eye, as
always with quantum theory things are not what they seem. Our
next Defense expert shows us how even the genius of von Neu-
mann may not be beyond reproach, as he manages to do the im-
possible: construct an interpretation of quantum theory
involving only ordinary objects. Let’s hear how this seemingly
impossible task was carried out.

THE QUANTUM POTENTIAL INTERPRETATION

Reality is undivided wholeness connected by “pilot waves”

In 1951 McCarthyism was running rampant across the Ameri-
can intellectual landscape, the father of the atomic bomb, J.
Robert Oppenheimer, being one of its prominent victims. At the
Atomic Energy Commission hearings against Oppenheimer, one
of the witnesses called was David Bohm, a young physics profes-
sor from Princeton, who had been one of Oppenheimer’s Ph.D.
students. Bohm refused to testify against his old professor, an
action that clearly irked the commission. Given the temper of
those times, such an act was tantamount to confessing to Com-
munist leanings of one’s own, and ways were found to strip

462

PARADIGMS LOST

Bohm of his professorial post. Following this clash with the au-
thorities, Bohm left the United States, finally settling in London
as professor of physics at Birkbeck College after leaving tempo-
rary havens in Brazil and Israel. Having brushed the dust of
McCarthyite witch-hunting reality off his boots, Bohm pro-
ceeded to the safer and infinitely saner and more rewarding con-
sideration of quantum reality. At this time he began to develop
an earlier idea of Louis de Broglie into a mathematically con-
sistent interpretation of quantum theory involving only ordi-
nary objects.

Interestingly enough, prior to his forced withdrawal from
teaching at Princeton, Bohm had authored what is still a highly
regarded textbook on quantum mechanics that follows the con-
ventional Copenhagen party line. But even though he was serv-
ing up this traditional Danish pastry to his students, Bohm was
becoming increasingly convinced through conversations with
Einstein that both Bohr and von Neumann were wrong — an or-
dinary-reality interpretation of quantum theory was possible.
The chink Bohm identified in von Neumann’s armor had to do
with an implicit assumption he made about the interaction of
quantum objects. Yon Neumann assumed that they interact in
what he termed “reasonable” ways. The kind of interactions
that Bohm had in mind would definitely not be “reasonable” by
von Neumann’s criteria, as we’ll soon see.

The key theoretical idea that Bohm based his approach upon
was the notion of a pilot wave. This idea had been introduced in
the 1920s by de Broglie but quickly laughed out of court by the
Copenhagenists in view of what looked to be insurmountable
mathematical difficulties. But Bohm showed how to overcome
those difficulties, reviving de Broglie’s idea of regarding a quan-
tum object as a particle with an associated pilot wave that, in
effect, tells it how to move. Let’s look at one or two of the details.

In the pilot wave picture, every quantum object is a real par-
ticle possessing definite attributes at all times. Associated with
each such object is a pilot wave that is also real but undetectable
other than through its effects on the particle. This wave is
termed the quantum potential, and serves the function of “read-
ing” the environment and reporting its findings back to the par-
ticle. Let me emphasize here that this is a real wave and should
not be confused with the quantum wave function, a purely mathe-
matical gadget for making predictions. The particle then acts in

HOW REAL IS THE ''REAL WORLD''?

463

accordance with the information provided by its associated pilot
wave. As a result, in the Quantum Potential Interpretation a
quantum object is not composed of a single “thing,” particle or
wave, but is both. Notice how objective reality is restored in this
picture, as there is no longer the ongoing schizophrenia between
the object as wave or particle. At all times it is both, and at all
times the particle side of the house possesses all the usual classic
attributes. B ohm’s genius was to show how this scheme could be
made to work. But there’s no free lunch in life or in quantum
theory either, and to those of a traditional outlook there is a
heavy price to be paid for this restoration of objectivity.

The first major complaint against the quantum potential ap-
proach is that it invokes a physically unobservable wave to re-
store objective reality. To most physicists, if you can’t measure
it, then it doesn’t exist, and the advantage of postulating an en-
tity such as the quantum potential is not worth the price of un-
detectability. When it comes to the trade-off between an
objective reality and physical observability, the verdict of the
working physicist is that without observability you have noth-
ing, or at least whatever you do have isn’t physics. But this ob-
jection to the quantum potential pales by comparison with the
other main argument against it: the need for faster-than-light
signaling.

It’s ironic to realize that the quantum potential was developed
as a step toward rescuing Einstein’s idea of reality from the
scrap heap where it was tossed by the work of Bohr and von
Neumann. Ironic because the biggest obstacle to the rescue oper-
ation is one that Einstein himself created by his Special Theory
of Relativity, with its ironclad prohibition against any kind of
signals being transmitted at a velocity faster than that of light.
Remember that the quantum potential acts something like a
radar wave sending out probes into the environment, with the
“reflected” signals used by the particle half of the quantum alli-
ance to decide what to do. Thus the quantum potential senses the
presence of a measuring apparatus of a certain type and imme-
diately notifies the particle, which then adjusts its behavior to
accommodate to the kind of attribute the device is designed to
measure. It can be shown that this kind of signaling from the
quantum potential back to the particle involves information
transfer of some sort that must move at superluminal rates — a
direct violation of Einstein’s cosmic speed limit.

464

PARADIGMS LOST

Bohm’s partial answer to this difficulty is that the quantum
potential is not a wave of matter, just a wave of active informa-
tion. Its effect depends only on its form, not upon its magnitude;
consequently, unlike matter waves such as sound or water whose
effect diminishes with distance from the source, the quantum po-
tential can have big effects at long distances. This is the phenom-
enon of nonlocality, which will occupy our attention shortly. In
Bohm’s view relativity is a statistical effect, not an absolute one.
The superluminal effect is seen only when we look at the correla-
tions between signals at two separated locations; if, however, we
look at what’s happening in the local neighborhood of either lo-
cation, the statistical properties of the signals appear to be inde-
pendent. Therefore no superluminal aspects show up.

In recent years Bohm has become an active spokesman for the
school of thought that sees the universe as a giant hologram,
arguing that to truly understand and be able to explain quan-
tum processes, we must abandon our traditional modes of reduc-
tionists thinking. Beneath the world of surface phenomena
there is an undivided seamless whole, and it is this “under-
world” that is the domain of quantum objects. In this realm,
every object is connected to every other because of the intertwin-
ing of their quantum potentials, ensuring that every quantum
object carries a trace of every other object with which it has ever
interacted.

With regard to the Measurement and Interpretation Prob-
lems, the Quantum Potential Interpretation comes off rather
well. It definitively resolves both problems in the way that Ein-
stein would have liked best (aside from the superluminal aspect,
which he most definitely didn’t like at all). Bohm’s theory deals
with the Measurement Problem in much the same way as the
Many-Worlds Interpretation: He says that the wave function
does not collapse because it doesn’t represent a complete descrip-
tion of the object. Once the additional variables are provided
(the quantum potential), there is no collapse, hence no Measure-
ment Problem. The Interpretation Problem is disposed of in a
similarly clean manner: All quantum objects are ordinary ob-
jects having all attributes at all times. Consequently, reality is
objective and independent of whether or not we happen to be
looking. Thus the Quantum Potential Interpretation solves
every problem that’s interesting about quantum reality — just as
long as you can accept “real” entities that are undetectable and

HOW REAL IS THE ''REAL WORLD''?

465

superluminal transfer of information. Before closing the book
on the Defense case, let’s hear from one more witness who has
revived an old idea of Wheeler and Feynman’s to provide yet
another interpretation that restores the objective reality of
quantum objects, but this time with a wave function collapse.

THE TRANSACTIONAL INTERPRETATION

Reality is a wave function traveling both backward
and forward in time

In our earlier story about the excited atom giving off its excess
energy to an atom of silver to darken a photographic plate, we
noted that prior to the process of absorption, the electron’s posi-
tion is described by a wave function that is created at the mo-
ment of its emission from the excited atom. In scientific
parlance, this is termed a retarded wave, for reasons that will
become apparent in a moment. Toward the end of the Second
World War, J. A. Wheeler and Richard Feynman proposed
what they termed an absorber theory of such emission processes
in which advanced waves are produced in such emission-absorp-
tion processes on an equal basis with retarded waves. The idea is
that when the retarded wave is absorbed by the atom of silver
(at some time in the future), there takes place a cancellation
erasing all traces of advanced waves and their effects. In this
theory, the silver atom absorber carries out its absorption of the
original retarded wave by manufacturing a second retarded
wave that is identical in amplitude but exactly out of phase with
the retarded wave from the emitting atom. In this way, the two
waves cancel and we speak about the original retarded wave as
being “absorbed.” In the Wheeler-Feynman theory, the silver
absorber also makes an advanced wave that “backtracks” the re-
tarded wave, moving backward in time along the path taken by
the retarded wave from the emitter. This advanced wave reaches
the emitter exactly at the instant of emission. It then continues
backward in time, but now is accompanied by the advanced wave
from the emitter. Since the two waves are exactly out of phase,
they also cancel, removing all “advanced” effects in the process.

When we observe this absorption of energy by the silver atom,
we don’t have access to these inner mechanisms of Nature. As a
result, all we see is that a retarded wave has traveled from the
excited atom to the photographic plate. Thus, from an observa-

466

PARADIGMS LOST

tional standpoint, the absorber theory leads to precisely the same
observations as any of the usual descriptions, e.g., the Copenha-
gen Interpretation. But the conceptual difference is considera-
ble, since now there has been a two-way exchange transferring
energy across spacetime from the excited emitter atom to the
absorbing silver atom. Recently, this idea has been taken up by
John Cramer of the University of Washington, forming the
heart of what he calls the Transactional Interpretation.

The key ingredient in Cramer’s view is that every quantum
event (interaction) involves such an advanced-retarded “hand-
shake” across space and time. This handshake is a sort of two-
way contract between the past and the future serving as the
vehicle for the transfer of energy, momentum, spin, and so
forth. While the details of Cramer’s arguments are a bit heavy
for our purposes here, the essential point is that the transaction
is explicitly nonlocal in the sense that the future is in some way
affecting the past, at least insofar as it enforces correlations be-
tween quantum events. For example, when we look through our
telescope at the light from the star Tau Ceti, which is eleven
light-years away, not only have the retarded light waves from
Tau Ceti been traveling for eleven years to reach our eyes, but
the advanced waves generated by the absorption processes
within our eyes have reached eleven years into the past, thereby
completing the transaction that permitted Tau Ceti to light up
our lives.

The great advantage of the Transactional Interpretation is
that it eliminates the observer from the formalism of quantum
mechanics. By this process, all of the paradoxes associated with
observer-dependent realities such as half -dead/half -alive cats,
waves of knowledge, and splitting universes vanish. The draw-
backs are that the vanishing act is performed by unobservable
phenomena (the advanced waves) transferring information and
energy at superluminal velocities. Note also that with the Trans-
actional Interpretation there is still a wave function collapse; in
fact, there are two collapses: one for the retarded wave, one for
its time-reversed counterpart. On the other hand, the naive-real-
ist requirement that quantum objects have well-defined proper-
ties at all times is retained. So like the Quantum Potential
Interpretation, the Transactional Interpretation involves real
entities that are unobservable and superluminal transfers of in-
formation. Swallowing the first is more a matter of taste than

HOW REAL IS THE ''REAL WORLD''?

467

experiment. Pursuit of the second takes us deep into the heart
of one of the most startling results in modern physics — the Bell
Interconnectedness Theorem. But before giving our attention to
this friend of the court, let’s pause to summarize the testimony
for the Defense in Table 7.4.

SCHOOL

WAVE FUNCTION
COLLAPSE

UNMEASURED

ATTRIBUTES

Naive Realist

always exist

Quantum Logic

always exist

Quantum Potential

always exist

Transactional

yes

always exist

TABLE 7.4. The dogwork realists

THE BELL TOLLS FOR LOCALITY

In his experiments on the paranormal, Duke psychologist Jo-
seph B. Rhine often employed card-matching experiments in
which a subject was shown the back of a Zener card whose front
might bear a star, cross, circle, square, or wavy lines. The sub-
ject was then asked to call out the pattern on the card being
shown, with significant deviations from the 20 percent random-
guess hit rate deemed evidence for ESP. Suppose we cook up a
variant of Rhine’s experiment to test for telepathy.

Our experiment will involve two subjects, Alexander and
Anastasia, placed on opposite sides of an opaque screen. Instead
of showing cards with patterns, the experimenter will show our
subjects questions randomly chosen from a fixed set of, say,
three possible queries, which are themselves randomly selected
from a supply of three-question sets. Further, assume that each
subject’s question is randomly selected on every trial. To keep
things simple, suppose the questions have only a simple yes or no
answer. Thus a specific set of questions could be: (1) “Do you
believe this experiment genuinely reveals anything about the ex-
istence of ESP?” (2) “Do you think the experimenter has any
idea of what he’s doing?” (3) “Are you doing this just for the
money?” Now imagine that the responses by Alexander and

468

PARADIGMS LOST

Anastasia repeatedly display this peculiar feature: Whenever
they are both shown a question bearing the same number they
both give the same answer. After several repetitions of the ex-
periment, always with the same result, the experimenter con-
cludes that the two are definitely in telepathic contact. He then
publishes papers in all the right ESP journals, claims a cushy
professorship, and is granted time on The Today Show to report
his astonishing findings to a world eager for scientific confirma-
tion of its deeply held beliefs about such matters. Is there any-
thing amiss here?

Despite public acclaim, fortune, film offers, and a cover ap-
pearance on People magazine, Alexander and Anastasia are
roundly denounced as fakes by impassioned scientists who claim
that the whole experiment is a fraud. The scientists point out
that the entire circus can be accounted for by the simple as-
sumption that the two are in communication before the experi-
ments begin. All they need to do is to agree in advance on what
their answers to the particular questions will be. Thereafter the
results are foreordained. For example, if they agree that they
will answer no to all questions numbered 1, yes to questions
numbered 2 and 3, there is no need for communication during
the experiment, and the “astonishing” outcome of their agree-
ment whenever the questions are the same is assured.

The less vocal (but more thoughtful) members of the intellec-
tual fringe note that not only are their more excitable colleagues
correct, but that the sort of agreement described is the only pro-
cedure by which the two fakers can possibly arrange things so
that the final outcome is always complete agreement on identi-
cally numbered questions. The claim of this less vocal, but far
more insightful, group of scientists is that it is not only suffi-
cient but also necessary that Alexander and Anastasia both
know the answer they will give to each question in order for
them always to concur whenever the question numbers are the
same. In addition, the thoughtful scientists have also kept track
of the answers given by the two “telepathists” when the question
numbers are ignored, i.e., when the pair are not told the number
of the question that they are being asked. These investigators
find that over a long run of experiments, the complete set of
answers given by the two differ exactly half the time, as would
be expected if the subjects just randomly answered yes or no. In
other words, if we just look at the string of answers from each

HOW REAL IS THE ''REAL WORLD''?

469

subject over a long sequence of experiments when they are unin-
formed as to the question numbers, the number of agreements
equals the number of disagreements, on the average. From this
fact they conclude that Alexander and Anastasia must have been
in some type of communication in those trials where they were
told the question numbers, trials that led to perfect agreement
when the question numbers were the same. Let’s see why.

We’ve already seen that the pair must have had some kind of
plan for how they would each answer the questions, for instance,
the pattern no-yes-yes stated above. Let’s abbreviate this to
NYY. Furthermore, if we observe that they always agree when
shown identically numbered questions, then our earlier argu-
ment demands that each uses the same plan. Our problem is to
show that Alexander and Anastasia must have actually com-
municated their respective plans to each other. To show this, first
of all assume that there was no communication. Under this as-
sumption, they should agree only half the time — on the average.
Now let’s compute the average number of hits and misses with
the NYY plan. Since there are three questions in each set, the
total number of cases is nine. These cases are shown in Table 7.5,
together with the answers given by the subjects using the NYY
plan.

It’s easily verified that any other plan involving two Y’s and
one N or two N’s and one Y will have the same result: five agree-
ments and four disagreements. Of course if the plan is NNN or
YYY, then there will be complete agreement. The upshot of this
entire business is that with any conceivable plan there are more

QUESTION

ASKED

(ALEXANDER

FIRST)

(1, 1)

(1, 2)

(1, 3)

(2, 1)

(2, 2)

(2, 3)

(3, 1)

(3, 2)

(3, 3)

Alexander’s

answer

Anastasia’s

answer

Agree/

disagree

TABLE 7.5. Experimental results with the plan NYY

470

PARADIGMS LOST

agreements than disagreements. This directly contradicts the
50-50 split seen when there is no communication. So we conclude
that Alexander and Anastasia are indeed telepathic.

While the saga of Alexander and Anastasia may seem rather
fanciful, it illustrates perfectly a crucial fact about the logic of
measurement processes. And when transferred to the quantum
domain, this experiment forms the basis for the Bell Intercon-
nectedness Theorem, a result that some have hailed as the most
profound discovery of science. Since Bell’s result was motivated
by von Neumann’s “proof” of the impossibility of hidden varia-
ble theories and Bohm’s subsequent “disproof,” the right place
to start our search for the essence of Bell’s message is to go back
to hidden variables and the notoriously puzzling Einstein-Po-
dolsky-Rosen (EPR) Paradox.

By now it should be evident that Einstein was always pro-
foundly unhappy with the idea of a statistical kind of quantum
world underlying the world of surface phenomena, spending the
last half of his life fighting an unrelenting guerrilla war against
the Copenhagenists and their anti-hidden-variable ontology. The
biggest salvo Einstein ever fired in this battle was the paradox
he concocted in 1935 with two colleagues, Boris Podolsky and
Nathan Rosen, designed to show that the quantum theory as
touted by the Copenhagenists provides only an incomplete de-
scription of Nature; i.e., there must be a more complete theory
involving variables hidden from the view of the current theory.
Let’s consider a simple version of the EPR experiment due to
David Bohm.

Imagine a device that generates pairs of electrons and shoots
them off in opposite directions. One of the attributes of an elec-
tron is its spin direction, the axis of which can point in one of
two directions. Let’s call them UP and DOWN. Since the total
spin in this two-particle system must be zero in order to con-
serve angular momentum, when one electron’s spin axis points
DOWN, the other must be UP, and conversely. EPR argued as
follows: Generate such a pair and let them separate so that one
remains here on Earth, with the other going to the great spiral
galaxy Andromeda 2 million light-years away. Now measure the
spin direction of the member of the pair that stayed here on
Earth. The Copenhagen view says that before the measurement
this electron had no definite spin at all. Rather it was in the state
“half UP, half DOWN,” just like the sorry state of

HOW REAL IS THE ''REAL WORLD''?

471

Schrodinger’s Cat. And similarly for its sibling over in An-
dromeda. But quantum theory states that as soon as the mea-
surement is made, voila: It’s TIP or DOWN with no ifs, ands, or
huts. Moreover, at that very moment the other electron in An-
dromeda somehow “knows” about the measurement here on
Earth and its spin direction is also definitely fixed DOWN or
UP, opposite to whatever was seen on Earth. The paradox is
now clear: How did that information get from here to An-
dromeda so fast? By Einstein’s own Special Theory of Relativ-
ity, it should have taken at least 2 million years for any kind of
signal to get to Andromeda. Yet according to the Copenhagen
position, the electron’s spin is somehow communicated instantly.
What’s going on here?

Naturally Einstein used the above thought experiment to
claim that the quantum description was deficient and that there
must be hidden variables whose values, if known, would have
given the earthbound electron and its Andromeda counterpart
definite and opposite spins at all times. In that case there would
be no need for the superluminal transfer of information to An-
dromeda, because the Andromeda electron wouldn’t need it to
know in which direction it should be spinning. In this way Ein-
stein tried to restore objective reality to the quantum world by
pitting the Copenhagen Interpretation against his own well-
tested Special Theory of Relativity. The quantum theorists and
Einstein batted the EPR ball back and forth across the net for
the better part of thirty years, with no definite resolution of the
point. The impasse was finally broken in 1964 with the publica-
tion in the first volume of an obscure physics journal of a six-
page paper using no more than elementary undergraduate
mathematics. That paper was by CERN physicist John Bell, and
served as the rallying cry for a whole new era of quantum real-
ity research.

The EPR argument rests upon the assumption that super-
luminal information transfer is impossible, hence quantum the-
ory cannot be complete. Bell took his cue from Bohm’s
demonstration that a hidden variable theory can make sense — at
least if information gets around faster than light — and discov-
ered that Einstein was wrong: Any viable hidden variable theory
must allow for faster-than-light communication. To dispel a pos-
sible misconception right at the outset, this does not imply that
you can send a message instantaneously to a friend in An-

472

PARADIGMS LOST

dromeda. Remember: Bell’s Theorem talks about the world of
deep reality, not the world of phenomena. We’ll return in a mo-
ment to a more detailed consideration of this point. First let’s
look a little more closely at Bell’s magnificent achievement.

To see the basic idea underlying Bell’s result, let’s go back to
our electron pair-generating device and assume that it regularly
shoots out electrons whose spin axes are randomly oriented. This
means that as each pair comes out, the chances are equal that the
spin axes of the pair point in any particular direction in space,
with, of course, one electron’s axis pointing opposite to its
twin’s. The general situation is shown in the figure below, where
the random spatial direction is p and the generator is denoted by
the symbol 0.

s «- ®

-P

Now suppose we have two identical spin-detection devices, one
on Earth and one on Andromeda, each with a direction knob al-
lowing the device to detect electrons spinning either in the direc-
tion of the knob setting or in the opposite direction, e.g., the
directions +p or ~p as above. Suppose at the outset that the
two devices are both set to the same direction — call it d. Since
detection of an electron is a yes/no proposition, let’s also agree
that each device records its observations on a tape, writing “1”
if it detects an electron, and “0” if nothing is detected. Since
initially both knobs are placed at the same directional setting, we
would expect the tapes from Earth and Andromeda to agree.
Thus a typical run might produce the following records:

Earth’s record: 01000101110011001101
Andromeda’s record: 01000101110011001101

Note the crucial point that although each sequence is random
(since the generator fires electrons whose spins are randomly
oriented), there is perfect correlation between the two records.
So although an observer on Earth and one on Andromeda would
see what looks like a purely random sequence of 0’s and l’s,
someone with access to both records would probably start draw-
ing conclusions about the two streams of electrons being related
in some nontrivial way.

Now let’s change the situation and set the knob on Earth at a

HOW REAL IS THE ''REAL WORLD''?

473

new direction, say d + 10°, leaving the Andromeda device alone.
In this situation, some of the electrons detected on Earth will
not be detected on Andromeda, and vice versa. Typical records
in this case might look like this:

Earth’s record: 011001(71110011011101
Andromeda’s record: 01(7001211100110(71101

Here the three mismatches are indicated in italics. Since there
are twenty trials in the run, when the offset between Earth and
Andromeda is 10° we have an error rate of 15 percent. Obvi-
ously, since the situations here and in Andromeda are perfectly
symmetrical, the error rate would have been exactly the same
had we left our knob alone and our friends in Andromeda been
the ones to make the 10° adjustment. Let’s denote this error rate
as EX 10°) = 15 percent.

To continue the experiment, suppose now that the Androme-
dans decide to twist their knob by 10° in the opposite direction;
i.e., their new setting is now d — 10°. Thus the relative difference
between Earth’s and Andromeda’s directions is now 20°. Big
question: What is the error rate E7(20°)? This question is easy to
answer if we assume that the errors on Earth are independent of
those in Andromeda; i.e., the errors we see when we use Earth’s
device as the standard are independent of the errors we see when
we use the Andromeda device as the benchmark. What this as-
sumption involves is the claim that whatever’s happening on
Earth has no bearing on what’s going on in Andromeda, and vice
versa. Under this working hypothesis, we can easily work out a
bound for the new error rate. Since the errors seen on Earth
were E(10°), we have to add to this the further errors intro-
duced by setting the Andromeda knob at d — 10°. By the sym-
metry of the situation, this error is also E7(10°). So we might
first conclude that the new error rate should be twice this
amount, i.e., E(2(f) = 2 EX 10°). But wait. When the Androme-
dans shifted their knob, we lost the standard for Earth’s record,
and similarly, when we twisted Earth’s knob, we destroyed the
standard for Andromeda’s record. Thus the overall effect is that
there will be some cases in which “double errors” cancel each
other out. That is, an error will be detected on both Earth and
Andromeda, each canceling the other’s effect so there appears to
be a match instead of a mismatch. Taking this factor into ac-
count, we can say only that the error rate of interest at 20° could

474

PARADIGMS LOST

not be greater than twice the rate at 10° but could possibly be
less. Symbolically, we can write this as 157(20°) < 2E( 10°). It
should be clear that the particular angles 10° and 20° have no
bearing on the argument, which is valid for any angle A . Thus
we can write E(2A ) < 2E(A ), which is the famous Bell in-
equality, the basis for Bell’s Theorem.

Now let’s look at the hypotheses we used in deriving the
foregoing result. There are two:

• Objectivity: We assumed that the electrons’ spin axes really
had a definite direction at all times between emission from the
generator and their measurements on Earth and in An-
dromeda. In other words, the electrons are ordinary objects.

• Locality: The errors seen on Earth and in Andromeda are com-
pletely independent of each other. In short, twisting the direc-
tion knob on Earth has no bearing on what’s seen in
Andromeda, and conversely.

By now you might be saying to yourself, “OK, this all looks
perfectly reasonable. What’s the point?” The point is that if you
actually perform this experiment with real electrons (but with-
out a confederate on Andromeda), you will find that the Bell
inequality is violated. In fact, according to the experiments of
John Clauser and more recently Alain Aspect, it’s violated to
such a degree that the possibility of attributing the deviation to
experimental error is negligible. In short, Bell’s result says that
either locality or objectivity (or, perhaps, both) has got to go!
Thus, if you want to keep an Einstein-Bohm type of hidden var-
iable reality, then you have to do as Bohm did and sacrifice lo-
cality. On the other hand, if you want to retain locality, which
most physicists of the Copenhagen stripe insist upon, then there
can be no hidden variables to bail out a naive-realist position.
And even if quantum theory ultimately turns out to be false,
Bell’s result will still hold: objectivity or locality, but not both.
So this fact, sometimes called Bell’s Interconnectedness Theo-
rem, places severe constraints on any pretender to the reality
throne, imposing the strict condition that if you’re advocating a
hidden variable approach and you haven’t explicitly included a
place for superluminal connections, then don’t bother submitting
your paper. Your theory cannot possibly be correct. No wonder
some have called this one of the most important results in the
history of physics.

HOW REAL IS THE ''REAL WORLD''?

475

* # #

The normal long-distance communication channels in America
used to be offered by Ma Bell; in the universe it seems that the
long lines have been laid down by Doc Bell. This is a good time
to reconsider the question of whether we could use this cosmic
Bell System to send a superluminal invitation for cocktails to
our colleagues in Andromeda. There’s been no small amount of
confusion on this point since Bell’s Theorem came out of the
physicists’ closet, an indicator being a flash letter from a Cali-
fornia think-tank executive to an Undersecretary for Something
at the Pentagon informing him of the result, and suggesting
that the ability to send such messages would offer an unjamma-
ble command-and-control communication system for submarines.
No doubt the second paragraph of the letter was an offer to look
deeper into the matter for a small consideration. Unfortunately
for that enterprising executive, the prospects appear distinctly
bleak for using Doc Bell’s channels for submarines or any other
kind of human or ETI contact. Let’s see why.

In our spinning-electron experiment, we saw that twisting the
knob on Earth had a definite effect on the correlation between
the record seen on Earth and that seen on Andromeda. Thus we
can definitely say that some kind of nonlocal effect was “caused”
on Andromeda by our action here on Earth. The problem is that
the Andromedans won’t notice anything unusual. All that will
happen is that they will get a record consisting of a different
random sequence of 0’s and l’s. Since they don’t know what re-
cord they would have received if we hadn’t twisted our knob,
there is no real transmission of information between us. The
only way that information could be transmitted would be if the
Andromedans had advance knowledge of the settings we were
going to be using. But there is still no superluminal method
known for transmitting this information. So, since one random
sequence looks pretty much like any other, Andromeda will have
no way of detecting the difference between two sequences attrib-
utable to different settings of our spin detector. And it is only in
such differences that a message can be coded. Thus, inspection of
the output of their detection device gives them no information
about the input to our device, because they don’t know the hid-
den variables (the setting of our device). So it seems that Doc
Bell’s communication channel is even better than the California
executive claimed: You can use the channel to send a signal

476

PARADIGMS LOST

that’s not only unjammable, but so perfectly scrambled that only
Nature holds the decoding key!

Despite his appearance here as a friend of the court, as long as
we have Bell on the stand it’s impossible to resist the temptation
of asking for his own position on the cases put forth by the
Prosecution and Defense. Given the choice of abandoning either
objective reality or locality, Bell casts his vote for retaining ob-
jective reality and the Quantum Potential Interpretation of
Bohm. Says Bell, “In my opinion, the pilot wave picture un-
doubtedly shows the best craftsmanship among the pictures we
have considered.” On this unambiguous note, let’s call in one
more friend of the court to tell us why information about the
earliest moments of the universe has some light to shed on how
reality really is.

IN THE BEGINNING, THE VERY BEGINNING

In 1964 two physicists from Bell Labs were trying to calibrate a
microwave communication antenna and found their efforts con-
tinually thwarted by some sort of background noise that they
were unable to account for by any Earth-based interference. The
ultimate explanation of that noise resulted in the award of the
1978 Nobel Prize in physics to the two researchers, Robert Wil-
son and Arno Penzias, for their discovery of “fossil evidence” of
nothing less than the moment of creation of the universe itself.
This discovery of the so-called microwave background radiation
was the final factor in tilting the scales between the Steady-State
Theory of the universe, which held that things have always been
more or less as they are today, and the Big Bang Theory, claim-
ing that the universe began in an explosion of literally cosmic
proportions. By consensus, the Wilson-Penzias noise is the elec-
tromagnetic residue of that primordial fireball and, along with
the observed expansion of the universe in all directions and the
abundance of the light elements — hydrogen, helium, and deute-
rium— it serves as the major selling point for the Big Bang The-
ory today.

If the Big Bang Theory is correct, the implication is that at
some time in the past, currently estimated at 12 billion years ago
(plus or minus a few hundred million), the universe was com-
pressed into a microscopic point of hard-to-believe proportions
and properties. What is absolutely astonishing is that physicists

HOW REAL IS THE ''REAL WORLD''?

477

now feel that they can give an almost letter-perfect account of
what happened after the first 10-30 second or so following the
universe’s birth. Just incredible! This amount of time is so short
that no kind of clock imaginable can even come close to measur-
ing it. Yet the large-scale structure of the universe as we see it
now is reasonably well explained by current theory after the
first 10~30 second. Unfortunately for the cosmologist as well as
the reality theorist, those first few ticks of the clock were where
most of the action took place, since that’s the time when what we
now call the laws of Nature were laid down and the structure
we see today was determined. To dig a little deeper into the
fundamental nature of these laws, let’s take a closer look at
what we do see today when we look at the large-scale struc-
ture of the universe, as well as some of the puzzling aspects
about it.

Two things you’d soon notice about today’s universe if you
looked at it through a powerful telescope is that it’s extraor-
dinarily homogeneous and isotropic (i.e., it is smooth and looks
the same in all directions). Thus, on the large scale the visible
matter has a remarkably uniform distribution; it is not orga-
nized into “clumps” separated by regions of empty space. Fur-
thermore, this is the picture you would observe regardless of the
direction in which you pointed your telescope. Besides the homo-
geneity and isotropy, after a few calculations you’d immediately
notice another peculiarity. The universe’s rate of expansion is a
bit like Goldilocks’s porridge: not too big and not too small, but
just right. So right, in fact, that a change of just a fraction of a
percent in one direction or the other in the force of gravity
would lead to an uninhabitable universe: either one in which
stars were born and died much too quickly for our kind of life to
evolve, or a universe in which matter could not have coalesced
into stars and galaxies at all. In short, the universe is “flat,”
precariously balanced on a knife edge between an open cosmos of
runaway expansion and a closed universe of rapid recollapse. In
view of the way things look today, the Big Bang might be com-
pared to a group of blindfolded football players gathered in a
huddle. The players are instructed to run away from the center
of the huddle in a straight line when they hear the referee’s gun
go off. The shot is fired and they start running, with the miracu-
lous result that the original huddle expands outward not in a
ragged* roughly circular fashion, but into an ever-growing per-
fect circle! This remarkable state of affairs cries out for an ex-

478

PARADIGMS LOST

planation, and somehow what was going on in that first 10~30
second holds the key. The first clue comes from what some call
number mysticism.

NUMEROLOGICAL PHYSICS

Homogeneity, isotropy, and flatness are not the only puzzling
coincidences we observe about the way the universe appears
today. There are also some oddly disturbing relationships be-
tween many of the basic constants that go to make up the so-
called laws of physics. In 1923 the British cosmologist Arthur
Eddington noticed a curious relationship between the gravita-
tional constant O, Planck’s constant h, the speed of light c, and
the mass of the proton mp. When he combined these basic con-
stants of Nature so as to cancel out their respective units of
measurement, thus obtaining a pure, dimensionless number, Ed-
dington found the following ratio:

~ 10”

What struck Eddington about this incredibly large number was
that, to within a factor of 10 or so, it is exactly the square root
of the number of protons in the universe, which is the immense
quantity Np ~ 1078 (here the symbol ~ means “approximately
equal”). Since there’s no a priori reason why the number of pro-
tons should bear such an uncanny relationship to the earlier
quantity, Eddington felt that he was on the track of a deep, un-
discovered principle of Nature, and invented an elaborate theory
to account for this type of numerical “coincidence.”

Later another eminent British physicist, Paul Dirac, followed
up Eddington’s ideas and discovered further remarkable rela-
tionships of the same sort linking the electric force e between the
proton and electron, the gravitational force between the same
two particles, the age of the universe tp, and the time needed for
light to cross an atom. Here are Dirac’s relations:

Electric force

~ 2.3 X 10”

Gravitational Force

Time for light to cross an atom

Age of the universe

HOW REAL IS THE ''REAL WORLD''?

479

To have these basic constants combine in such a way as to arrive at
virtually the same outrageously large number was just a little heav-
ier dose of coincidence than Dirac was prepared to accept. Thus,
he made the bold assertion that the two ratios were in fact identi-
cal, leading (after a minor amount of algebra) to the estimate:

mpmlcl

(*)

This relation for the age of the universe is written to show that
the only part of it that is not fixed is the term involving the
gravitational constant G. The rest of the terms involve masses,
the speed of light, and the like, all of which are assumed to be
unchanging over the course of time. But since the age of the
universe is obviously not time-invariant, Dirac concluded that in
the relationship he had discovered, the gravitational coupling
constant is steadily decreasing as time goes on, to keep things in
balance. Furthermore, from Eddington’s relations we see this
has the consequence that the number of protons, in the universe
must be increasing with the square of the age of the universe,
implying that matter is continually being created.

At the time Dirac made these claims in the late 1930s, they
caused a small stir in the cosmology community. However, later
experiments using the Viking lander to measure the orbital pe-
riod of Mars showed that Dirac’s idea of a time-varying G is
very unlikely to be correct, since the period was not changing as
would be required if G were not constant. So are these “coinci-
dences” just coincidences, or is there still a real explanation
lurking in the wings? In 1961 Robert Dicke of Princeton pub-
lished an argument asserting that Dirac’s coincidence was no co-
incidence, not at least if one accepts what has now come to be
termed the Weak Anthropic Principle. Since Dicke ’s anthropic
argument has formed the basis for more than a little contro-
versy in the physics community over the years, it’s worth our
time to devote a few pages to a deeper consideration of its basis
and conclusions.

ANTHROPIC PRINCIPLES

At the most uncontroversial level, anthropic reasoning comes
down to the well-accepted principle that when you’re engaged in
measuring anything, it’s necessary to take into account the par-
ticular properties of the measuring instrument. When the in-

480

PARADIGMS LOST

struments happen to be ourselves as human beings, then the con-
clusions from our measurements have to respect the peculiar
features giving rise to our situation as observers. And the most
important such features are the physical conditions that appear
to be necessary for our very existence at this time, on the third
planet circling a typical G-type star in the suburbs of the Milky
Way Galaxy. This idea, in effect, is what underlies the so-called
Weak Anthropic Principle (WAP), which can be stated as:

weak anthropic principle: The observed values of all physical
quantities are restricted by the requirement that they be compati-
ble with our existence as observers.

The reader will recognize this kind of reasoning as a middle
ground between the pre-Copernican view of mankind as the cen-
ter of the universe, and the post-Copernican cosmology, which
denies mankind any special status or position. The Weak An-
thropic Principle states, in effect, that while our position may
not be central, it is privileged to some degree.

In his 1961 paper, Dicke employed the WAP as an explana-
tion for Dirac’s numerical relations. His argument is instruc-
tive. On the basis of well-known principles of nuclear physics,
Dicke calculated that the expression on the right side of relation
(*) on page 479 should very closely approximate the lifetime of
a typical star. So it’s not at all surprising, he claimed, that these
same constants will combine to equal roughly the age of the uni-
verse. Reason? Since the matter from which we are constructed
must have first been synthesized in the nuclear reactions at the
core of a star, the universe cannot be younger than the lifetime
of a star, or we would not be here to worry about the question.
End of proof. Since it’s crucial for understanding the heated
debate between the proponents and opponents of anthropic argu-
ments, note carefully the chain of reasoning in Dicke’s argu-
ment:

1. Given the existence of mankind, the age of the universe could
not have a value much different from the one it actually has.

2. Thus, Dirac’s relations don’t apply to any universe, but only
to the universe that we actually observe today.

This pattern of logic completely reverses the direction of rea-
soning usually employed in science. Generally we start by speci-
fying the initial situation and the laws of Nature, then predict

HOW REAL IS THE ''REAL WORLD''?

481

the subsequent state of affairs. Anthropic reasoning proceeds in
the opposite direction: Start from the final observed state (now),
and try to constrain the initial situation by asserting that it
could only have been one that would give rise to a universe that’s
inhabited today by intelligent observers like ourselves. So
Dicke’s technique is to cite a present condition (our existence) as
an explanation for a phenomenon having its source in the past
(the age of the universe). Up to this point — the introduction of
intelligent observers — most physicists will at least grudgingly
accept the tenets of the WAP, even if many of them do believe it
smacks more of tautology than principle. But there is a stronger
version of the principle, termed (not very imaginatively) the
Strong Anthropic Principle (SAP), which makes intelligence the
key actor in the cosmic drama.

The WAP says nothing about the laws of physics themselves,
nor does it comment on the actual values of the fundamental
constants like the speed of light or the gravitational coupling
constant. It simply tries to explain various observed features of
the universe, taking these items as givens. The SAP, on the
other hand, tries to use anthropic reasoning to attach actual val-
ues to these quantities. An example will illustrate.

Suppose the gravitational constant G were a million times
larger than it actually is. Then the lifetime of a star in its life-
giving phase would be about a million times less, since the higher
gravitational forces would greatly accelerate the burning of its
nuclear fuel. But even in such a universe Dicke’s argument
would still apply. If an observer exists in such a universe, when
the age of that universe is around ten thousand years he would
see a universe whose mass would be a trillion times smaller than
ours. Question: Would life arise in such a vastly accelerated uni-
verse? The WAP is totally silent on this issue; the SAP says no,
life can exist only if the fundamental constants have values very
close to their observed levels.

The foregoing sort of argument leads to the most familiar
form of the SAP:

strong anthropic principle: The universe must be nearly as we
know it or life would not exist; conversely, if life didn’t exist, nei-
ther would the universe.

The reader will immediately note that the gap separating the
SAP from the classic argument from design invoking a super-

482

PARADIGMS LOST

natural Creator is no more than a hairsbreadth, omitting only
an explicit invocation of a Designer. Finally, for the sake of
completeness, let’s note the so-called Final Anthropic Principle
(FAP), which asserts the kind of ultimate fate for intelligence
that virtually all traditional religions would endorse, namely,
that our descendants will become like gods.

final anthropic principle: Once life is created, it will endure for-
ever, become infinitely knowledgeable, and ultimately mold the

universe to its will.

If this kind of argument sounds familiar it should, since it’s
central to the Austin Interpretation of J. A. Wheeler considered
earlier. Wheeler argues that for a universe to be real, it must
evolve in such a way that observers come into existence. One of
the main pillars supporting his contention is what he calls the
Participatory Anthropic Principle (PAP), asserting that the uni-
verse is brought into existence by the collective observations of
all intelligent observers who have ever existed or who ever will
exist. At about this point, skeptics like Martin Gardner start
trotting out principles of their own, like the Completely Ridicu-
lous Anthropic Principle (CRAP), as antidotes to the high-fly-
ing assertions of these “anthropic physicists.” As a micro-
example of a typical academic feud, let’s look at a few of the
arguments for and against the anthropic principles to see how
they may or may not help us come closer to understanding the
true nature of Nature.

Heinz Pagels of Rockefeller University was one of the most
vocal opponents to the use of anthropic principles in physics
prior to his untimely death in a rock-climbing accident in 1988.
Pagels claimed that anthropic principles are “the lazy man’s ap-
proach to science.” He cited at least three main deficiencies in
the use of such reasoning in the practice of science, arguing that
anthropic principles: (1) use the unknown to explain the known;
(2) never predict anything and are entirely post hoc; (3) are both
immune to experimental falsification and untestable. Pagels con-
cluded his indictment of the “anthropicists” by saying that the
anthropic principles are “the closest that some atheists can get
to God.” It’s amusing to note that in a popular article on the
topic arguing in favor of anthropic reasoning, the physicist Tony
Rothman used similar words: “When confronted with the order

HOW REAL IS THE ''REAL WORLD''?

483

and beauty of the universe and the strange coincidences of na-
ture, it’s very tempting to take the leap of faith from science
into religion. I am sure many physicists want to. I only wish
they would admit it.”

As to the claims that the anthropic principles are untestable,
unfalsifiable, and post hoc, supporters point out that Dicke could
have used the WAP to argue against the Steady-State Theory
of the universe even before the observations of Wilson and
Penzias. The argument is that in the Big Bang Theory the age
of the universe happens to be approximately equal to 1 /H,
where H is the expansion rate of the universe. However, in the
Steady-State Theory, H must by definition be constant and
therefore has nothing to do with the age of the universe. Conse-
quently, in the Steady-State Theory there’s no reason for \/H to
equal the lifetime of a typical star. So the fact that it does is
either a gigantic coincidence or an anthropic argument in favor
of the Big Bang. The fact that Dicke did not make this claim in
no way argues against the inherent possibility of generating a
testable prediction on the basis of the WAP. To see another kind
of prediction that can be obtained using anthropic arguments,
let’s return to the topic of the last chapter and quickly reprise
Brandon Carter’s anthropic-based argument against the exis-
tence of ETI.

As we know, those arguing in favor of the likelihood of ETI
often invoke the Principle of Mediocrity to buttress their claims.
What this comes down to is a special case of the Copernican
Principle asserting that there’s nothing special about life on
Earth. Consequently, since the universe is so vast, and since we
are here, the chances are great that “they” are there. Wheeler,
for one, counters with the anthropic claim that the universe is
vast only because it’s several billion years old, and needs to be
that old to give rise to one intelligent civilization (ours). Since
there’s no particular reason why there should be ETIs out there,
additional civilizations would be wasteful of the universe’s re-
sources. This line of reasoning has been vastly sharpened by
Carter, who sparked off the current “anthromania” in a 1974
address to the International Astronomical Union in which he
coined the term “anthropic principle.” We have already given
the essence of Carter’s argument against ETI in the last chap-
ter, so there’s no need to repeat any more than the basic idea
here.

484

PARADIGMS LOST

First of all, Carter assumes that there are a number of indi-
vidually improbable steps on the road to intelligent life. Next he
predicts the average time between the emergence of an intelli-
gent species and its death from, say, the burning out of its sun.
Finally, he argues (on the basis of the WAP) that intelligent
life is exceedingly rare. So we conclude that if ETIs are found
with any frequency, Carter’s WAP-based prediction is wrong.
This constitutes a testable prediction using anthropic argu-
ments: Just find lots of ETIs out there, and Carter’s WAP-
based argument will be falsified. But we’re starting to wander
off course from our original goal of looking at the quantum cos-
mological doings in the first 10-30 second of the universe’s exis-
tence. So let’s close this short excursion into anthropic thinking
with the following food for thought from Freeman Dyson: “As
we look out into the Universe and identify the many accidents of
physics and astronomy that have worked together to our benefit,
it almost seems as if the Universe must in some sense have
known that we were coming.”

QUANTUM COSMOLOGY

Since in the Big Bang picture the universe was much smaller
than an atom in the very early going, we have to use the con-
cepts of quantum theory to describe what was happening in
those first few picopico . . . picoseconds. You might object that it
seems to go well beyond the bounds of credulity to imagine that
the whole universe could be compressed into a volume far less
than that of an atom, since the energy density must have been
intolerably large. But remember, according to quantum theory
energy and time are conjugate variables, so we can get large
amounts of energy into a small volume if the time is short
enough. If 10~30 second isn’t short enough for you, perhaps
you’ll need to seek your fortune on the astral plane after all. In
any case, let’s stay in this universe and look at some of the ex-
planations for how the large-scale features of the universe could
have emerged out of this “point” of matter-energy.

The two main puzzles surrounding the moment of the Big
Bang center on the seemingly highly ordered nature of the fly-
speck of matter-energy constituting the initial state, and the ex-
traordinarily delicate balance in the gravitational force that left
our universe teetering right on the edge between a runaway ex-

HOW REAL IS THE ''REAL WORLD''?

485

pansion and a too-rapid collapse. Let’s talk here only about the
Initial State Paradox.

The crux of the paradox is that the observed homogeneity and
isotropy of the universe today, not to mention the conditions
necessary for our existence, are difficult to account for by any-
thing other than a highly ordered initial state. Yet if we choose
initial states of the universe randomly, the chances are over-
whelmingly high that the state that pops up will be very disor-
dered. The situation here is exactly the same as that faced when
we deal out the cards in a game of poker. According to the prob-
ability theorists, there are a total of 2,598,960 possible initial
hands that could turn up on the deal in a round of five-card
draw poker. Let’s assume that these possibilities stand for the
various possible initial states of the universe. Now let’s ran-
domly dip into the deck and select a hand for our universe. Ac-
cording to current theory, the chances of getting a hand that
corresponds to an initial state favorable to our type of life, and
that is consistent with the kind of large-scale structure we ob-
serve, are vastly less than the likelihood of having a royal flush
staring you in the face when you pick up your poker hand. And
this probability is only 4 in 2,598,960, or a bit better than 1 in a
million! So how can we account for the apparently highly un-
likely initial state of our universe? Several answers have been
proposed.

•Many-Universes Theory: This resolution of the paradox is the
cosmologist’s appeal to Everett’s Many-Worlds Interpretation
in quantum theory. The Everett theory postulates a different
branch of the universe for each possible value of an observable
quantity, so what could be more natural than to claim that our
universe just happens to be one of the few in which all the condi-
tions and constants came out “just right”? Note here the appeal
to the WAP as a self -selection mechanism for choosing a “good”
universe for life from the set of possibilities, almost all of them
“bad.”

Because of its neat disposition of the Initial State Paradox,
the Many-Worlds Interpretation is a favorite among cosmolo-
gists. In fact, among all the quantum interpretations considered
earlier, Everett’s is the only one that really gives a consistent
and coherent picture of how to deal with the initial state prob-
lem. However, opponents argue that it is the very antithesis of

486

PARADIGMS LOST

Ockham’s Razor, being far too extravagant in dispensing “uni-
verses for all occasions” to be taken seriously as a solution to the
dilemma.

•Dissipation: Adherents to this view claim that the initial state
was not so well ordered at all, but that frictional and other dissi-
pative forces smoothed out the initial inhomogeneities. Thus tur-
bulent mixing and recombination of the primordial matter soon
led to the kind of regular state we see today. Opponents argue
that if we admit disordered initial states, there are always some
such states that are so nonuniform that even after billions of
years the irregularities would not have been dissipated. Further-
more, as we know from rubbing two rough surfaces together,
friction generates heat, and calculations show that the amount
of dissipation needed to arrive at today’s universe would have
generated an amount of heat far in excess of what’s observed in
the Wilson-Penzias background radiation. So at the moment dis-
sipation isn’t seen as a likely solution to the paradox.

•Inflation: At present the main scientific opponent to the Many-
Universes Theory is the idea that the early universe enjoyed a
short inflationary period, which smoothed out the initial state;
thereafter, the universe settled into its current expansionary
mode. This is a little bit like what happens when you blow up a
balloon. Initially the balloon has no air and is just an irregular,
crinkly rubber sack. However, as soon as you pump in the first
couple of breaths of air, the balloon immediately springs into a
smooth, regular shape, which expands uniformly thereafter.

The inflationary model, originally proposed by Ed Tryon in
the early 1970s and later developed by Alan Guth at MIT, postu-
lates a repulsive force that operated against gravitational forces
to expand the universe to about the size of a basketball during
the first 10-35 second after the Big Bang. At this point the re-
pulsive primeval force split into the four forces we know about
today (gravitational, electromagnetic, weak nuclear, and strong
nuclear), and the familiar radiation-dominant force of expan-
sion took over. An important feature of this scenario is that it
allows the universe to have come into existence as nothing more
than a quantum fluctuation in a total vacuum. The matter
needed for Nature to pull off this conjuring trick came, of
course, out of Einstein’s famous formula E = me2, which shows

HOW REAL IS THE ''REAL WORLD''?

487

the equivalence between matter and the energy contained in the
vacuum. In short, everything comes out of nothing!

At present inflationary models seem to have the upper hand in
the cosmological derby, although those committed to a more an-
thropic view note that one can give an anthropic explanation for
why the universe is so isotropic that’s just as convincing as the
one obtained by invoking inflation. For example, Hawking and
Collins have argued (via the WAP) that if the universe were not
isotropic, then we wouldn’t be here to observe it. Carrying this
argument one step further, they would claim that the initial
state must have been special, too. Opponents would (and do) say
that while there’s nothing wrong with this line of reasoning, it’s
certainly not necessary, and that an idea like inflation is aes-
thetically more satisfying. On this note, let’s move to the final
contender for the solution of the Initial State Paradox.

•God: This is clearly the most straightforward solution of all.
Simply invoke a Grand Designer who stirred up Goldilocks’s
porridge to exactly the right temperature and consistency so
that both the initial state and the fundamental constants of Na-
ture came out “just right” for us to be here. This is the familiar
argument from design, which has been the mainstay of all non-
scientific accounts of the universe from time immemorial and
needs no further amplification here.

As a postscript to the quantum cosmology issue and the Initial
State Paradox, it’s amusing to consider for a moment the final
state from the anthropic point of view. If we believe in the Final
Anthropic Principle, there might not be much to choose between
the argument from design and the idea that our successors will
ultimately come to be indistinguishable from God. This argu-
ment follows from Wheeler’s Participatory Anthropic Principle,
which requires intelligent life to have a significant effect upon
the large-scale properties of the universe. Following up the im-
plications of the FAP, many scientists and philosophers have
come to the conclusion that if life evolves in all the many uni-
verses in a quantum cosmology, and if life continues to exist in
all these many worlds, then all of these universes will approach
what the French Jesuit priest and mystic Pierre Teilhard de
Chardin called the Omega Point. As noted by the anthropicists
Frank Tipler and John Barrow:

488

PARADIGMS LOST

At that moment, life will have gained control of all matter and
forces not only in a single universe, but in all universes whose
existence is logically possible; life will have spread into all spatial
regions in all universes which could logically exist, and will have
stored an infinite amount of information, including all bits of
knowledge which it is possible to know. And this is the end.

And this is the end for us, too, in our account of the anthropic
principles and their possible relevance for the problem of reality.
Let’s now give the floor back to the lawyers for their final argu-
ments.

SUMMARY ARGUMENTS

Both the romantic realists and the dogwork realists have argued
extensively and persuasively to convince us of the rightness of
their respective causes. Before summarizing the positions, let’s
again review the issue before the house. Put simply, we have the
Prosecution’s claim:

There is no such thing as a unique, observer-independent reality.

On the other side of the courtroom, we hear the Defense say,
“Maybe not.” At least it says there is no irrefutable evidence to
conclude that an objective deep reality, independent of observ-
ers, does not underlie the world of phenomena. Tables 7.6 and
7.7 summarize the competition.

Before I enter into a justification for my own conclusion on
this ultimate question, let me pull an ace from up my sleeve and
say that whatever position you care to hold, the experimental
data will not refute you. As it turns out, each of the above posi-
tions is in complete accord with the experimental evidence! So
until there’s an experimental breakthrough of some kind, the po-
sition you hold on the quantum reality issue is more like a reli-
gious conviction than a matter of science. All positions are
defensible, and your choice becomes as much a matter of aesthet-
ics and a gut feeling for “how it could be that way” as a logical
consequence of hard facts. With this extraordinary situation in
mind, allow me to close out this all-too-brief tour of life, behav-
ior, cognition, language, machines, and universes with my pri-
vate prejudices as to the reality of reality.

HOW REAL IS THE ''REAL WORLD"?

489

THERE IS NO OBJECTIVE REALITY!

PROMOTER

ARGUMENT

Bohr (Copenhagen

Interpretation)

overall measurement situation

von Neumann, Wigner
(Consciousness

Interpretation)

consciousness determines reality

Wheeler (Austin Interpretation)

measurement option

Heisenberg (Duplex
Interpretation)

potentia and actuality

Everett, Deutsch (Many Worlds
Interpretation)

every world is a reality

TABLE 7.6. Summary arguments for the Prosecution

A SINGLE, OBSERVER-INDEPENDENT
REALITY MAY EXIST!

PROMOTER

ARGUMENT

Einstein (naive realist)

Newtonian reality is real

von Neumann, Finkelstein
(quantum logic)

nondistributive logic

Bohm, Bell (quantum potential)

pilot wave theory

Cramer (transactional events)

advanced and retarded waves

TABLE 7.7. Summary arguments for the Defense

BRINGING IN THE VERDICT

The paradox of the quantum realm is that although common
sense dictates that the universe exists “out there” independent
of acts of observation, the universe does not actually seem to
exist “out there” independent of acts of observation. One view is
that we are insignificant specks playing out totally uneventful
roles in a vast cosmic play; the alternate position says that in
some way we are not only the players, but the drama’s writer,
director, and producer, as well as critic and audience, too. It’s

490

PARADIGMS LOST

hard to be of more central importance than that! As I’ve tried to
cut this Gordian knot of conflicting scientific visions of reality,
my own oscillations between the arguments of the Prosecution
and Defense have come to symbolize for me the essence of the
dilemma itself: “How can it possibly be that way?” In the final
analysis perhaps we all think of ourselves as romantics at heart,
so my personal struggle with the nature of reality comes to a
temporary halt with a vote for the Prosecution and its clients,
the romantic realists. Specifically, I give the nod to Everett’s
Many -Worlds Interpretation.

When it comes to sifting the evidence and claims, as I men-
tioned above the experimental evidence really offers no help. Ev-
erything known from the laboratories is perfectly consistent
with the MWI and any of the other contending views. So it ulti-
mately comes down to a matter of aesthetics, and to my mind at
least, the MWI just has a few more selling points than the com-
petition. To begin with, it has fewer ad hoc assumptions, espe-
cially about the mysterious measurement act. That the physical
act of attaching a Geiger counter, camera, microscope, or meter
stick to some system should dramatically affect the basic nature
of things is still a difficult notion for me to swallow. The MWI
manages a fairly clean resolution of this problem by the simple
expedient of denying that there is any problem. Second, the
MWI appears to be the only quantum reality that gives a coher-
ent picture of the Initial State Paradox of cosmology. Since the
way we see the laws and state of the universe today is condi-
tioned by the character of that initial state, an interpretation
that gives some kind of scientifically defensible account, even if
it does seem bizarre, looks better to me than the scientific equiva-
lent of a Gallic shrug or an even more outlandish explanation.
Finally, there is Bell’s Interconnectedness Theorem. Such a re-
sult cannot be proved in the MWI for obvious reasons: The
proof relies on the fact that while many outcomes of a measure-
ment are possible, only one of them is actually realized; i.e., we
need a counterfactual condition to prove the result. In a cosmos
where all possible outcomes are realized, there is no Bell’s Theo-
rem. To my mind, banishing this kind of superluminal connec-
tion is a definite plus for the MWI. Of course, any reality in
which happenings around Procyon or over in Andromeda affect
earthly doings has plenty of nonlocality of its own, even without
Bell’s result. Nevertheless, I feel more comfortable with this

HOW REAL IS THE ''REAL WORLD''?

491

kind of nonlocality than with the Bell type.

Before closing I should say a word or two about the dogwork
realists, in particular the quantum potential crowd. When I first
learned about quantum mechanics and started pondering the
fateful question, I naively wondered why it wasn’t possible to
regard an electron simply as a particle that moved along its ap-
pointed Newtonian path in wavelike fashion, with a continual
back-and-forth “wavy” type of locomotion like that of a fish or a
snake. While my ignorant musings were hopelessly adrift in a
technical sense, they don’t seem that far away in spirit from
what’s presented in the quantum potential, or pilot wave, pic-
ture. The view of a quantum object as a particle with an as-
sociated wave appears to me to be only one step (albeit a gigantic
one, conceptually) removed from my early vision. So when it
came time to vote in the reality game, I was sorely tempted to
cast my ballot for the quantum potential. But when all was said
and done, as a romantic at heart I just couldn’t resist a roman-
tic reality, and the MWI is far and away the most romantic of
them all. So while my mind is with the quantum potential, my
heart is with the MWI. And so is my vote.

CONCLUSION

THE BALANCE
SHEET

ARE HUMANS REALLY
SOMETHING SPECIAL?

WHERE DO WE STAND?

Physicists and philosophers love principles: the Heisenberg Un-
certainty Principle, the Principle of Conservation of Energy,
the Principle of Parsimony (Ockham’s Razor), Fermat’s Princi-
ple, and many more. For centuries one of the most inviolable of
all such principles was Aristotle’s Principle of Continuity, by
which Nature passes gradually from the most imperfect forms
here on Earth up to the most perfect works of God in heaven.
By this reckoning, hell was at the center of the Earth, hence the
center of the universe. A natural corollary of this principle is

CONCLUSION

493

that mankind occupies a central position in the universal scheme
of things. Later Copernicus displaced mankind from its unique
position in the most dramatic fashion possible. With his Coper-
nican Principle, he argued that no one part of the universe is
more privileged than any other. The Principle of Continuity and
the Copernican Principle represent the antipodes of the human
role in the universe: mankind at the heart of all things versus
mankind as an insignificant speck on the cosmic horizon. We are
living at one of those rare moments in which the pendulum is
swinging through its midpoint, on its way back to the human-
centered universe of Aristotle. And our age has its own princi-
ple, the Anthropic Principle, asserting man’s role as the measure
of all things. In one way or another, all of our stories in this
book have been accounts of what science has to say about this
anthropocentric claim. So as prelude to a summing-up, let’s
reexamine our multiple foci.

The Big Question serving as the leitmotiv of our journey
through the jungles of modern science can be simply stated as:

Is there anything special about human beings?

Each of the stories I’ve told in traversing the uncharted terrain
of science has addressed this Big Question from its own particu-
lar vantage point: human biochemical structure, behavioral pat-
terns, cognitive capacities, and so on. Some of the stories relate
to the uniqueness of humans here on Earth; others deal with our
role in the galaxy, or even the universe at large. And as we pass
from one of these venues to another, the precise form of the Big
Question varies accordingly. Yet the overall theme has always
remained the same: Are we unique in any way that really
counts? To come to a verdict on this question, let’s briefly revisit
each of our topical areas and rephrase the Big Question in terms
suitable for illumination by that area’s special sort of lamp. In
this way, perhaps, a few glimmerings of our “specialness” may
emerge from these individual pieces of evidence.

Origin of Life. In this chapter we considered our material struc-
ture, the particular carbon-based biochemical processes by which
all known life forms on Earth operate. A version of the Big
Question appropriate to this context would be: Is the particular
way in which life arose here on Earth a statistical fluke, unlikely
to be repeated anywhere ever again? Or is the combination of

494

PARADIGMS LOST

steps leading to Earth’s life forms an almost inevitable outcome
given similar environmental conditions?

To be alive, any object must somehow possess the capability
for metabolizing raw materials from its environment into prod-
ucts needed to maintain itself. Moreover, the object must also be
capable of some kind of self-repair of its metabolic and repro-
ductive machinery, as well as production of copies of itself, per-
haps in conjunction with other members of its species. In our
consideration of these matters, we saw that on the basis of gen-
eral theoretical arguments by von Neumann and others, any
such object must possess structures that perform certain dis-
tinct functional activities: a constructor, a controller, a copier,
and so forth. So we could regard the Big Question in this setting
as being tantamount to asking: How likely is it that organisms
would arise elsewhere that possessed these functional capabili-
ties as part of their physicochemical makeup?

On the basis of the various explanations put forward for the
emergence of life here on Earth, my impression is that should
the Earth be wiped clean of all life today in some kind of plane-
tary Armageddon, the likelihood of life forms of any kind ree-
merging in a few billion years would be a bet that not even
Lloyd’s of London would put on the board. Consequently, ori-
gin-of-life considerations suggest to me that there is indeed
something special not only about humans, but about life in gen-
eral, as we see it here on Earth today.

Sociobiology. Moving from biochemical structure, we next ex-
amined the degree to which human behavioral activities, espe-
cially those of a social nature, somehow distinguish us from the
animals. In particular, we were concerned with whether these
behavioral traits were primarily determined by genetic program-
ming or, alternately, were principally a product of environmen-
tal (read: cultural) considerations. In this instance, a good
phrasing of the Big Question might be: Are most human social
behavior patterns innate, or are they primarily acquired by
means of learning and/or cultural conditioning?

In our examination of this question, the arguments flew fast
and furious. Relevant aspects of biology, genetics, and sociology
were mixed with politics and ideology in an ever-shifting blend
of logic, experiment, and raw emotionalism. While there was
considerable evidence to support the claim that many higher ani-

CONCLUSION

495

mals behave as if they are following the dictates of their genes,
the gap between these animals and Homo sapiens is a large one,
and one that the more vocal opponents of the sociobiologists as-
sert will never be bridged.

After all the rhetoric, smoke, and ashes drift away, my feeling
is that the sociobiological debate offers the least clear-cut evi-
dence one way or the other on the Big Question. Even at this
point, about the best I can offer is the opinion that human be-
havioral repertoires could very well be special, differing in essen-
tial ways from the basically genetic determination of other
living things. For me the sociobiological verdict still comes out
as nothing more conclusive than a definite maybe.

Language Acquisition. General social behavior is one thing; the
specific behavioral trait of spoken language is something else al-
together. This area took us into a consideration of whether lan-
guage capacity is part of the genetic birthright of every human
being. Or is language a human skill that’s acquired along with a
variety of others as part of a general learning capacity? Here
our Big Question comes down to asking: Is the human language
acquisition capacity a unique product of the way that the human
brain and body happen to be put together? Or can it be expected
to occur in any sufficiently complex organism capable of general
probing, learning, and problem solving in its environment?

Of all the evidence put forth in this book for the uniqueness of
humans, in my view the language acquisition case is by far the
strongest. The Chomskian assertion that there is a language ac-
quisition device that’s part of our genetic makeup seems far and
away a more convincing explanation for the observed facts
about language acquisition than any of the countertheories of-
fered by either the behavioral or cognitive psychologists. While
the neurophysiological evidence for the location, or even exis-
tence, of this device may still be far from conclusive, my gut
feeling is that the day is not far off when the boundaries of the
language device will be precisely determined and Chomsky’s po-
sition vindicated. Thus my view is that the language acquisition
evidence points strongly toward the position that a human being
is indeed a pretty queer bird.

Artificial Intelligence. Closely related to the language acquisi-
tion problem is the general question: Is it possible, in principle,

496

PARADIGMS LOST

to construct a machine that displays the same kind of cognitive
powers as a human being and, moreover, carries out these cogni-
tive tasks in the same way? When translated into these terms,
our Big Question becomes: Is there anything unique about our
way of thinking? Or, more specifically, can we duplicate human
cognitive processes in a machine?

Some, like Wittgenstein, have argued that there is no distin-
guishable difference between our language and our thinking.
While I’m far from convinced of the validity of this claim, at
least in any strong sense, even a weak form of it immediately
suggests close connections between the “thinking machine” ques-
tion and the problem of language acquisition. In our considera-
tion of the AI problem, a lot of philosophical arguments were
put forward showing why a machine could never think like you
and me. On the other side of the field stand the computer scien-
tists and engineers arguing that the final score should not be
tallied when the game has barely begun.

Strangely enough, while I feel that the language evidence
points clearly to our special nature, here I find myself siding, at
least provisionally, with the computer scientists and engineers.
Thus, on the basis of the AI evidence, I might conclude that our
cognitive capacities are not so special after all. How can I ex-
plain this clear-cut contradiction to my earlier position vis-a-vis
language? Basically, I can’t. My best effort is to argue that the
language problem indicates that humans are special as compared
with all other living agents here on Earth. But computers are
not living agents (at least not yet), and I find no essential con-
tradiction in thinking that perhaps a genuine thinking machine
is yet a possibility. Anyway, I’m afraid I must come up with a
negative reading on the Big Question here.

Extraterrestrial Intelligence. Moving away from Earth, our first
stop was the Milky Way Galaxy and the question of whether
there are other living, intelligent beings out there for us to com-
municate with. Here we might phrase the Big Question in the
form: As living, intelligent, communicating entities, are human
beings unique in the galaxy?

Using the Principle of Mediocrity, a corollary of the Coperni-
can Principle, astronomers gave us arguments showing why the
galaxy should be teeming with ETIs. On the other hand, we ex-
amined a number of biological, physical, and anthropic argu-

CONCLUSION

497

ments indicating that the chances for the existence of an ETI
are vanishingly small, essentially zero. Unhappily, I find the lat-
ter category of pessimistic arguments far more convincing than
those of the optimists, leading to the sad conclusion that we
probably are alone, at least in the galaxy. And, in fact, if the
universe is finite, the same arguments seem to point to the even
more disturbing conclusion that we could very likely be alone in
the universe as well. So on the strength of these ETI considera-
tions, humans again start looking like something very special in-
deed.

Quantum Reality. The final stop on our stroll through the won-
derland of science was nothing less than the universe of phenom-
ena with its accompanying puzzler: What is the nature of the
deep reality underlying observed phenomena? In particular, we
examined the role of humans as observer/participants in the cre-
ation of the underlying “stuff” from which the world of phe-
nomena is built. Here we could pose the Big Question in the
form: Is a human presence necessary to bring reality into exis-
tence?

Most of the group we termed the romantic realists gave argu-
ments suggesting that there is no such thing as an objective
physical reality, independent of human observers. The opposi-
tion, led by Einstein, argued otherwise. On the basis of the ac-
tual experimental evidence, we saw that there are no grounds for
accepting either side’s case as the last word. Nevertheless, a va-
riety of aesthetic considerations make it at least plausible, if not
desirable, to lean toward the romantics, thereby thrusting man-
kind into the role of creator as well as observer and participant.

In one last attempt to bring everything together, Table 8.1
summarizes my overall impressions on the Big Question from
each of the foregoing perspectives.

To my eye, the overall conclusion is that homo sapiens is a very
special creature, at least here on Earth, and maybe in the uni-
verse as a whole. While it may not yet be a conclusion to bet
your pension on, I think the odds favoring our uniqueness are
high enough that my bookie would tell me, “Off the board, doc.”
Since it would take a volume nearly the size of this one to ad-
dress adequately the many implications of this conclusion, let me
close this brief survey of science and the nature of mankind by
mentioning just one of them.

498

PARADIGMS LOST

ARE HUMANS SPECIAL?

origin of life

probably

sociobiology

hard to say

language acquisition

very likely

artificial intelligence

maybe not

extraterrestrial intelligence

very probably

quantum reality

arguably yes

TABLE 8.1. The bottom line

In his scathing indictment of the modern American university
in the recent bestseller The Closing of the American Mind, Allan
Bloom notes with alarm the gradual transformation of the uni-
versity from a community of scholars providing a liberal arts
education to what one of my colleagues has described as “a trade
school for the bewildered.” An important count in Bloom’s in-
dictment is the disappearance of any systematic study of the
great works of literature, philosophy, and the arts from the pro-
gram of today’s undergraduate, an observation that goes hand
in hand with the increasing illiteracy rate in the population at
large. Bloom, a humanist, sees the problem from the vantage
point of the college of liberal arts; many of the same signs also
appear in the college of science and engineering. As a longtime
habitue of this corner of the campus, I too have noted with
alarm an ever-accelerating trend toward more and more special-
ized courses and programs of the trade-school variety, neces-
sitating elimination of broader perspectives on the domains of
science and their many interrelations. My conclusion is that it’s
not just the concept of a classical liberal arts education that’s
endangered; it’s the very notion of education itself, liberal arts
or otherwise.

An important part of Bloom’s solution to the problem is a re-
turn to the Great Books: Plato, Shakespeare, Tolstoy, & Co. In
the same spirit I would advocate a return to the Great Problems
to reverse the trend toward fragmentation and incoherence in
the sciences. And in my view, the problem areas we’ve covered in
this volume — the origin of life, quantum reality, sociobiology,
and all the rest — are certainly prime candidates for inclusion on
anybody’s list of Great Problems. These problems share the
same virtue as the Great Books: They force one to learn about
the mutual interrelationships of many things. For example, it’s

CONCLUSION

499

inconceivable to me that anyone could even begin to address the
origin-of-life question without a good knowledge of chemistry,
molecular biology, evolutionary biology, and, most likely, com-
binatorics and computer modeling as well. In the same vein, con-
tributing to the AI question requires an understanding of
mathematical logic, the theory of computation, cognitive psy-
chology, neurophysiology, computer engineering, and program-
ming languages. Similar remarks could be made for the other
topics we’ve looked at in this volume. The point is not even that
the Great Problems are solvable by these means, but rather that
there’s so much to learn about the overall landscape of science
and the different ways of scientific thinking by expanding our
horizons and going beyond narrow, traditional, intradisciplinary
thinking.

To close on a somewhat somber note, Table 8.1 appears to lead
to the verdict that there is truly something special about hu-
mans. Nuclear holocaust, cosmic catastrophe, AIDS, and a thou-
sand other demons sit waiting to snuff out this small flicker of
intelligence and light in what looks like a vast, empty void.
Whatever we humans are and whatever we can be, I think we
have a responsibility not to lose it through carelessness and
neglect, benign or otherwise. Periodic reflection on the assess-
ments given here may help us keep this imperative in mind.
Let’s hope so.

TO DIG DEEPER

CHAPTER ONE
WORLD VIEWS IN COLLISION

The story of Jocelyn Bell and the discovery of pulsars is surely one of the
more exciting episodes in science during the turbulent 1960s. A firsthand ac-
count of the work by the lady herself is given in
Wade, N., “Discovery of Pulsars: A Graduate Student’s Story,” Science,
189 (1975), 358-364. '

See also the interview with Bell in the volume
Judson, H. The Search for Solutions. New York: Holt, Rinehart and Winston,
1980.

The first account of pulsars as rapidly rotating neutron stars appears to
have been given by Thomas Gold at a 1968 symposium at the International
Centre for Theoretical Physics in Trieste, Italy. The precise citation is

TO DIG DEEPER

501

Gold, T. “The Nature of Pulsars: Survey of Present Views,” in Contempo-
rary Physics, Trieste Symposium 1968, Vol. 1, pp. 477-481. Trieste, Italy: In-
ternational Centre for Theoretical Physics, 1969.

A detailed, scholarly account of the pros and cons of the entire Vaffaire Veli-
kovs ky is given in the highly enlightening work

Bauer, H. Beyond Velikovsky. Urbana, IL: University of Illinois Press,
1984.

This work is notable not only for its thorough investigation of the scientific
basis of Velikovsky’s claims, but also the detailed discussion of the form and
content of the criticism Velikovsky received. All in all, the author concludes
that while Velikovsky was very likely wrong from the standpoint of science,
it’s not possible to prove him wrong. Furthermore, the critics were themselves
far from beyond reproach, at least insofar as the methods they employed. In
this connection, Bauer cites the impassioned criticism by astronomer Carl
Sagan, who got so carried away in his denunciation of Velikovsky that he
ended up using the unconsciously held belief that science offers certainty and
truth, the creed of “scientism.” I highly recommend this book as a demonstra-
tion of how modern science operates when wearing both its logical and sociolog-
ical hats.

However, Bauer himself is not beyond using some of the same rhetorical
tricks he accuses Velikovsky’s acolytes of employing. For a sympathetic, nev-
ertheless critical view of Bauer’s book, see
Gardner, M. The New Age: Notes of a Fringe Watcher, pp. 65-71. Buffalo, NY:
Prometheus, 1988.

For an account of the ideas that got the whole Velikovsky business off and
running, see the “source”:

Velikovsky, I. Worlds in Collision. New York: Doubleday, 1950,

DID YOU SAY SCIENCE?

The common perception of science is as a means for “gadget production”; a
collection of facts leading to practical ends. But scientists think of science as a
set of methods and conceptual schemes leading to an understanding of natural
processes. For an informative, educational, and easily readable discussion of
this profound misunderstanding, see

McCain, G., and E. Segal. The Game of Science, 4th Edition. Monterey, CA:
Brooks/Cole, 1982.

The conventional ideology of science is an amalgam of the views of philoso-
phers, historians, and sociologists about the logic, progress and norms of the
scientific process. It is presented in digestible form in

Broad, W., and N. Wade. Betrayers of the Truth: Fraud and Deceit in Science.
New York: Simon and Schuster, 1982.

The foregoing book is especially notable for its detailed discussion of the
“missing link” in the conventional ideology — the human factor. The authors
conclude that the very nature of the ideology increases the attractions of
fraudulent activity in science, as well as the likelihood that such activity will

502

TO DIG DEEPER

go undetected. The authors claim that the root cause of the problem is that the
system based upon the conventional ideology ends up rewarding not only genu-
ine success, but also the appearance of success. As an account of the dark side
of science that many in the scientific establishment go to great pains to pooh-
pooh, this book is hard to beat.

THE NATURAL PHILOSOPHER'S STONES
A first-rate discussion of all the difficulties associated with the use of induc-
tion, as well as details on the various “-isms,” see
Chalmers, A. What Is This Thing Called Science f, 2nd Edition. Milton
Keynes, UK: Open University Press, 1982.

For a gentle introduction to philosophical problems arising in connection with
science, the following work is highly recommended:

Kemeny, J. A Philosopher Looks at Science. Princeton, NJ: Van Nostrand,
1959.

Wittgenstein once wrote that he thought it would be possible to write a seri-
ous work on philosophy that consisted entirely of jokes. His idea was that if
you understood the joke, then you would get the philosophical message. John
Allen Paulos took this dictum seriously, producing the following extremely en-
tertaining, as well as informative, work from which I shamelessly lifted the
little joke in the text on the Problem of Induction:

Paulos, J. I Think, Therefore I Laugh: An Alternative Approach to Philosophy.
New York: Columbia University Press, 1985.

For a detailed, even relentless, pursuit of the diagram for mathematical
modeling displayed in Figure 1.2, see the volume
Rosen, R. Anticipatory Systems. Oxford: Pergamon, 1985.

A novel work that attempts to explore the nature of reality as seen from the
perspectives of literature, sociology, physics, art, film, and a variety of other
fields is

Exploring Reality, D. Cohn-Sherbok and M. Irwin, eds. London: Allen and
Unwin, 1987.

The quote by Kalman relating to the instrumentalist view of the world is
taken from

Kalman, R. “Identification from Real Data,” in Current Developments in the
Interface: Economics, Econometrics, Mathematics, M. Hazewinkel and A. Rin-
nooy Kan, eds., pp. 161-196. Dordrecht, Netherlands: Reidel, 1982.

This paper, as well as several others noted in its bibliography, presents a par-
ticularly graphic portrayal of an attitude commonly held by many so-called
hard scientists: If you can’t measure it, it doesn’t exist — literally! Happily, as
time goes by such unimaginative and increasingly indefensible prejudices are
being weeded out of the scientific mindset, to be replaced by far less precise,
but vastly more enlightening, perspectives.

RATIONALITY FOR REALISTS

The text discussion of the work of Wittgenstein, Popper, et al. is nothing more
than a caricature of their deep, insightful views on the theory of knowledge,

TO DIG DEEPER

503

language, science, and reality. Two of the best general references to appear in
years on the interplay between the ideas of these philosophers and the logical
workings of science are

Newton-Smith, W. The Rationality of Science. London: Routledge and Kegan
Paul, 1981.

Oldroyd, D. The Arch of Knowledge. New York: Methuen, 1986.

A wonderful picture of the entire political, psychological and sociological
climate in Austro-Hungary leading up to the views of the Vienna Circle and
much, much more is offered in the volume
Johnston, W. The Austrian Mind. Berkeley, CA: University of California
Press, 1972.

Another work purporting to cover somewhat the same territory, and one that
has received enormous amounts of (in my opinion) undeserved publicity, is
Janik, A., and S. Toulmin. Wittgenstein’s Vienna. New York: Simon and
Schuster, 1973.

A personal survey taken through the years I’ve lived in Vienna shows that of
seventeen friends who’ve started reading this paralyzingly dull volume, not one
has gotten further than the middle of Chapter 3. Frankly, the only thing I can
see that the book has going for it is a catchy title, which undoubtedly accounts
for its continuing sale to unsuspecting tourists in the Viennese bookshops. My
recommendation: Stick with Johnston unless, of course, you’re looking for in-
stant insomnia relief.

In addition to the general philosophical sources cited below, a firsthand ac-
count of the discussions of the Vienna Circle and their relationship to the work
of Wittgenstein is provided in

Ludwig Wittgenstein and the Vienna Circle: Conversations Recorded by Friedrich
Waismann, B. McGuiness, ed. Oxford: Basil Blackwell, 1979.

A good, short biography of Wittgenstein is
Pears, D. Wittgenstein. Glasgow: William Collins and Sons, 1971.

Amusingly, given his later stance, Popper was initially attracted to Marxism
in his youth and spent some time working as a laborer. He later renounced
these leftist views, and has subsequently placed great emphasis upon the im-
portance of democratic principles. A good sample of the various philosophical
and social ideas of Popper can be found in the collection
A Pocket Popper, D. Miller, ed. London: Fontana, 1983.

For a personal account by Popper himself of the evolution of his views, see his
autobiography :

Popper, K. Unended Quest: An Intellectual Autobiography. Glasgow: William
Collins and Sons, 1976.

Lakatos died in 1974 at the relatively young age of fifty-two. As a result,
much of his work was published posthumously. For a summary of this work
and its significance, see

Essays in Memory of Imri Lakatos, R. Cohen, et al., eds. Dordrecht,
Netherlands: Reidel, 1976.

Feyerabend, P. “Imr6 Lakatos.” British Journal for the Philosophy of
Science, 26 (1975), 1-18.

504

TO DIG DEEPER

Lakatos s own statement of his idea of a Scientific Research Program is pre-
sented in his classic essays

Lakatos, I. The Methodology of Scientific Research Programmes. Cambridge:

Cambridge University Press, 1978.

Lakatos, I. Proofs and Refutations. Cambridge: Cambridge University Press

1976.

In considering the matter of public debate, it’s of interest to note Feyera-
bend’s description of his experiences as a young student attending the well-
known Alpbach symposia: “I met outstanding scholars, artists, politicians and
I owe my academic career to the friendly help of some of them. I also began
suspecting that what counts in a public debate are not arguments but certain
ways of presenting one’s case. To test the suspicion I intervened in the debates
defending absurd views with great assurance. I was consumed by fear — after
all, I was just a student surrounded by bigshots — but having once attended an
acting school I proved the case to my own satisfaction.” By his own admission,
Feyerabend not only comprehended a useful social truth, but also used it to lay
the basis for his later intellectual eccentricities, some of which are recounted in
his famous work

Feyerabend, P. Against Method: Outlines of an Anarchistic Theory of Knowl-
edge. London: New Left Press, 1975.

As an aside, the Dadaist movement promoted a somewhat sacrilegeous, irrever-
ent attitude toward art, with nothing to be taken seriously. It is exactly this
kind of attitude that Feyerabend advocates for the philosophy of science.
When portrayed in this light, perhaps his ideas aren’t so outlandish, after all.
Unfortunately for Feyerabend and the rest of the “sociology of knowledge”
theorists, it’s difficult to point to even a single form of a physical relation that
was determined by the social order or structure in which it was formed.

BUDDY, CAN YOU PARADIGM?

The story about Julian Bigelow, von Neumann, and “nobody’s” dog, as well as
much background information about Thomas Kuhn, is presented in the im-
mensely entertaining history of the geniuses and eccentrics of the Princeton
Institute for Advanced Study:

Regis, E. Who Got Einstein’s Office 1 Reading, MA: Addison-Wesley, 1987.

Ironically, Kuhn’s pathbreaking work on paradigms in science appeared in
The International Encyclopedia of Unified Science, a series of books from the Uni-
versity of Chicago Press that was an outgrowth of the logical positivist move-
ment led by Rudolf Carnap when it moved to America during World War II.
The precise citation is

Kuhn, T. The Structure of Scientific Revolutions, 2nd Edition. Chicago: Uni-
versity of Chicago Press, 1970.

This edition contains a lengthy postscript by Kuhn in which he addresses many
of the critical remarks leveled at the ideas in the original edition of 1962.

Those readers looking for a somewhat gentler introduction to the paradigm
notion, without having to wade through the customary dense prose of histori-
ans and philosophers, should see

TO DIG DEEPER

505

Briggs, J., and D. Peat. The Looking Glass Universe. New York: Cornerstone
Library, 1984.

This little masterpiece of scientific exposition addresses not only some of the
basic epistemological issues raised by Popper, Kuhn, and others, but also
treats a variety of the more speculative and exciting paradigms in the scientific
world today. Among the items considered are Prigogine’s theory of far-from-
equilibrium systems, Bohm’s ideas on language and quantum theory, and Shel-
drake’s theory of morphogenetic fields in developmental biology. All in all, one
of the best volumes available to give the general reader a glimpse into some of
the edges of today’s frontiers of science — and thought!

The heart of Shapere’s continuing critique of Kuhn’s position is found in his
review of the first edition of Kuhn’s book, which appeared as
Shapere, D. “The Paradigm Concept.” Science, 172 (1971), 706-709.

PHILOSOPHICALLY SPEAKING

For an outstanding reference on the ways replication and induction are carried
out in scientific practice, see

Collins, H. Changing Order. London: Sage Publications, 1985.

This volume is particularly important for its in-depth consideration of how the
Problem of Induction is solved in a sociological, or practical, sense in everyday
science. The author details the mechanics of this procedure by considering
three case studies in physics, engineering, and psychology: the detection of
gravitational radiation, the construction of an infrared laser, and the emo-
tional life of plants. Throughout, the author presents a lucid, enlightening and
entertaining summary of the interactions between the philosophical difficulties
of induction and the practical means by which science goes about dealing with
them. Highly recommended.

A TALE OF TWO SUICIDES

An account of Boltzmann’s suicide set against the general social and intellec-
tual climate of turn-of-the-century Vienna is given in the Johnston book cited
earlier. Kammerer’s life and tragic death are recounted with great detail and
sympathy in

Koestler, A. The Case of the Midwife Toad. London: Hutchinson, 1971.

A less detailed account of the Kammerer case, told within the general context
of fraud in science, is given in the Broad and Wade volume cited earlier.

The original form of Merton’s norms can be found in his classic work
Merton, R. K. The Sociology of Science. Chicago: University of Chicago Press,
1973.

Other excellent accounts of the sociology of science easily accessible to the gen-
eral reader are

Richards, S. Philosophy and Sociology of Science: An Introduction, 2nd Edition.
Oxford: Blackwell, 1987.

Ziman, J. An Introduction to Science Studies: Philosophical and Social Aspects of
Science and Technology. Cambridge: Cambridge University Press, 1984.

506

TO DIG DEEPER

While the above treatments look at the practice of science from the sociologi-
cal standpoint, an alternate approach is to look upon the whole enterprise from
the perspective of an anthropologist. In this view we regard scientists as if
they were members of some strange, hitherto unknown tribe, who spend their
days practicing arcane and mystical rites. The job is to understand the struc-
ture, language, customs and so forth of this “tribe” by using the commonly
accepted concepts, methods, and procedures of cultural anthropology research-
ers. A fascinating account of an experiment of just this type involving biologi-
cal work at the famed Salk Institute is given in

Latour, B., and S. Woolgar. Laboratory Life: The Construction of Scientific
Facts, 2nd Edition. Princeton, NJ: Princeton University Press, 1986.

A short account of the Summerlin incident is given in the Broad and Wade
book noted above. For all the gory details, the interested reader should consult
Hixson, J. The Patchwork Mouse. New York: Doubleday, 1976.

ON THE FRINGE OR AT THE CUTTING EDGE?

Two classic treatments of monkey business masquerading as science are the
volumes

Gardner, M. Fads and Fallacies in the Name of Science. New York: Dover,
1957.

Gardner, M. Science: Good, Bad and Bogus. Buffalo, NY: Prometheus Books,

1981.

However, my own tastes lean more toward the outstanding discussion given in
Radner, D., and M. Radner. Science and Unreason. Belmont, CA: Wadsworth,

1982,

from which the list of pseudoscience “fingerprints” in the text is taken.

THE PULPIT AND THE LAB

The story of Mrs. Fernandez and her “trial by prayer” is recounted in
Raup, D. The Nemesis Affair. New York: Norton, 1986.

This book gives a participant’s account of one of the most heated of today’s
scientific controversies, the problem of what happened to the dinosaurs. How-
ever, the author uses this issue as a vehicle to discuss much more general and
far-reaching questions about belief systems in science and the role they play in
shaping what a particular community comes to think of as “good work.” Thus,
the book serves as an admirable attempt to explain the evolution of a paradigm
crisis as it unfolds in real time.

On the matter of belief systems in science, Raup thinks most scientists would
claim that science involves the use of experiments to test hypotheses and care-
ful scholarship with no prior commitment to a particular answer. Also, he feels
they would argue that religion is not science because it involves no experi-
ments, tests no hypotheses, and is committed beforehand to a set of beliefs.
Raup says that these scientists’ claims contain a lot of bunk. In other words,
the ideal of science as broadcast far and wide by the PR division of the scien-
tific establishment, and the actual practice of science as carried on down at the
lab bench, bear little if any resemblance to one another. Just as I’ve always

TO DIG DEEPER

507

suspected! This little confession by Raup calls to mind the remark by Austin
when informed that Godel had shown that there were truths of arithmetic that
could not be derived from the Peano axioms. Austin remarked, “Whoever
thought otherwise?”

The interplay of observations, laws, theories, and models, not only in science
but also in religion, is covered nicely in

Barbour, I. Myths, Models, and Paradigms: A Comparative Study in Science and
Religion. New York: Harper and Row, 1974.

This book is to be recommended not only for its comparative analysis of the
scientific and religious enterprises, but also for much worthwhile background
information about the role of myths in the process of reality generation. Of
special note is Barbour’s discussion of the uses of models in religion, where
he offers the following competing visions of the relationship between God
and man:

God =

monarchical — king and kingdom

deistic — clockmaker and clock

dialogic — one person and another person

agent — agent and his actions

social process — individual and community

Another volume covering some of the same ground, but with a slightly more
biased stance, is

Hummel, C. The Galileo Connection: Resolving Conflicts Between Science and the
Bible. Downer’s Grove, IL: InterVarsity Press, 1986.

An excellent volume giving not only an overview of the quasi-religious charac-
ter of much of science, but also a general audience introduction to a spectrum
of questions, problems, and proposed solutions in science is
Stableford, B. The Mysteries of Modem Science. London: Routledge and
Kegan Paul, 1977.

INTO THE COURTROOM OF BELIEFS
The quote from Bauer is taken from his book on Velikovsky cited above.

CHAPTER TWO
GENERAL REFERENCES

Since Oparin, the origin of life has been a topic of continuing fascination for
scientists and the lay public alike. In recent years there have been several ex-
cellent treatments for the general reader. Two that were instrumental in shap-
ing my own view of the field are

Scott, A. The Creation of Life: Past, Future, Alien. Oxford: Blackwell, 1986.
Shapiro, R. Origins: A Skeptic’s Guide to the Creation of Life on Earth. New
York: Summit, 1986.

508

TO DIG DEEPER

The Scott book is notable for an excellent account of the biochemical aspects of
life, as well as for a truly first-rate collection of diagrams and figures illustrat-
ing some of the trickier points in the way life works. The Shapiro volume,
while not illustrated, is highly recommended as an account of the competing
positions from a skeptical, but not hostile, point of view.

A slightly more technical presentation of the “facts of life” is the following
textbook account aimed at university undergraduates:

Day, W . Genesis on Planet Earth: The Search for Lifers Beginnings , 2nd Edi-
tion. New Haven, CT: Yale University Press, 1984.

OUT OF THE FIRE AND INTO THE SOUP
By today’s standards the details of Oparin’s program for the origin of life, if
not the direction, seem hopelessly adrift. But the importance of his work for a
scientifically based attack on the problem cannot be overemphasized. To illus-
trate the type of “scientific” view as opposed to religious dogma held prior to
Oparin, one need only recount the “spontaneous generation” ideas of the Flem-
ish chemist and physician Jan Baptista van Helmont, who gave the recipe:
“Dirty undergarments encrusted in wheat; twenty-one days is the critical pe-
riod. The mice that jump out are neither weanlings nor sucklings, but fully
formed.” While Pasteur stamped paid to this ridiculous idea in the nineteenth
century, it was not until the work of Oparin that a serious scientific attack on
the origins question was mounted. As an entertaining aside on the spontaneous
generation hypothesis, despite Pasteur’s work the theory didn’t finally expire
until its last bastion, the British scientist Henry C. Bastian, died. Regrettably,
this seems part of the typical life cycle of discredited theories. In any case, the
original work of Oparin can be found in the following English version:

Oparin, A. Origin of Life, S. Morgulis, trans. New York: Macmillan, 1938.

New York: Dover reprint, 1965.

The independent proposal of Haldane, which gave rise to the term “primor-
dial soup,” is found in the essay

Haldane, J.B.S. “The Origin of Life,” in On Being the Right Size and Other

Essays. Oxford: Oxford University Press, 1985.

An account of Oparin’s political activity during the Lysenko period is given
in the Shapiro book noted earlier. See also the definitive account of the whole
deplorable Lysenko affair given in

Medvedev, Z. The Rise and Fall of T. D. Lysenko. New York: Columbia Uni-
versity Press, 1969.

Miller’s personal account of the circumstances surrounding his classic exper-
iment appears in

Miller, S. “The First Laboratory Synthesis of Organic Compounds Under

Primitive Conditions,” in The Heritage of Copernicus , J. Neyman, ed., pp.

228-241. Cambridge, MA: MIT Press, 1974.

In connection with Miller’s experimental parameters, it’s worth taking note of
the fact that the initial run of the experiment produced nothing of interest.
Only when Miller interchanged the order of the spark discharge and the con-
densing chamber did measurable amounts of any kind of amino acids emerge.

TO DIG DEEPER

509

This is a point worth pondering in regard to the problem of unacceptable in-
vestigator interference with prebiotic simulations; evidently, the problem was
there from the very outset.

As an indication of the faith that Cyril Ponnamperuma places in Nature’s
ability to generate amino acids from simple chemicals, as a side activity he is
chairman of the council of the Dambala Institute, a center devoted to exploita-
tion of the dambala plant (a kind of winged bean) as a protein source to solve
the Third World hunger problem. Ponnamperuma states, however, that this
is only an interim solution, his ultimate goal being to generate proteins di-
rectly from primitive elements in the atmosphere (carbon, nitrogen, hydrogen,
and so on). He thinks we could make up to 20 percent of our food that way,’
the principal limitation being the energy required for the synthesis. As to
where he stands on the origin of life on Earth, his statement “If I can demon-
strate a replicating molecule, I’ll die a happy man” says it all. For a more
complete account of his ideas on the food problem, as well as on prebiotic syn-
thesis, see

“Seeds of Life: An Interview with Cyril Ponnamperuma.” Omni, 1980.

Additional references to Ponnamperuma ’s work are given later under Chapter
Six, devoted to the existence of extraterrestrial intelligence.

A CRASH COURSE ON HOW LIFE LIVES
Simple and entertaining general accounts of the mechanisms of life include the
Scott book cited earlier, as well as

Hofstadter, D. ‘The Genetic Code: Arbitrary?” in Metamagical Themas, pp.

671-699. New York: Basic, 1985.

Rosenfield, I., E. Ziff, and B. Van Loon. DNA for Beginners. London: Writ-
ers and Readers Publishing, 1983.

A somewhat more technical presentation of the facts is

Rose, S. The Chemistry of Life, 2nd Edition. London: Penguin, 1979.

Those not convinced that a self-reproducing factory is possible are urged to
read the prologue of the book

Hogan, J. P. Code of the Lifemaker. New York: Ballantine, 1983.

POTHOLES ON THE ROAD TO LIFE
The “junk” DNA problem has recently been attacked by computer simulation
experiments run in Material Mode by Loomis and Gilpin at UC, San Diego.
They speculated that much of the excess DNA is just there by chance. Using a
simulation program embodying various rules for DNA replication, they found
that a single gene will blossom into a genome containing many genes, some of
which are members of multigene families, and all of which are embedded in a
very large proportion of dispensable sequences. Hence, they conclude that: (1)
most of the DNA in eukaryotic genomes does nothing at all, and (2) large
quantities of dispensable sequences will accumulate in the genome before it will
stabilize. An account of their work is found in

Loomis, W., and M. Gilpin. “Multigene Families and Vestigial Sequences.”

Proceedings of the National Academy of Sciences USA, 83 (1986), 2143.

510

TO DIG DEEPER

A summary of these experiments is

Lewin, R. “Computer Genome Is Full of Junk DNA.” Science, 232 (1986),
577-578.

The WEES simulator idea is discussed in great detail in the paper
Lahav, N. “The Synthesis of Primitive ‘Living’ Forms: Definitions, Goals,
Strategies and Evolution Synthesizers.” Origins of Life, 16 (1985-86)
129-149.

MONSTERS, HYPERCYCLES, AND NAKED GENIES
A detailed description of the background, experimental setup, and results of
Spiegelman’s pioneering experiment is given in
Spiegelman, S. “An in Vitro Analysis of a Replicating Molecule.” American
Scientist, 55 (1967), 3-68.

The complementary experiment by Eigen is presented in
Eigen, M., W. Gardiner, P. Schuster, and R. Winkler-Oswatitch. “The Ori-
gin of Genetic Information.” Scientific American, 244 (1981), 88-118.

A fairly complete description of Orgel’s ideas about creating template-di-
rected RNA without benefit of enzymes is presented in
Orgel, L. “The Origin of Life and the Evolution of Macromolecules.” Folia
Biologica, 29 (1983), 65-77.

The Gilbert scenario for the origin of life out of self-catalytic RNA is out-
lined in

Gilbert, W. “The RNA World.” Nature, 319 (1986), 618.

On the puzzle of junk DNA, Gilbert’s view is that it arises as the relic of the
old intron-exon structure left imprinted on the DNA from the RNA molecules
that originally encoded proteins.

A complete expository and technical account of much of the “hypercycle”
theory underlying the ideas of Eigen is found in
Eigen, M., and P. Schuster. The Hypercycle: A Principle of Natural Self-Orga-
nization. Berlin: Springer, 1979.

See also the 1981 Scientific American article cited earlier.

The computer experiments of Niessert are reported in

Niessert, U. “How Many Genes to Start With? A Computer Simulation

About the Origin of Life.” Origins of Life, 17 (1987), 155-169.

Niessert, U., D. Harnasch, and C. Bresch. “Origin of Life Between Scylla
and Charybdis.” Journal of Molecular Evolution, 17 (1981), 348-353.

A spectrum of theories beyond those discussed in the text have also been
offered to explain why nucleotides came first. Perhaps the most intriguing is
the hydrated electron theory of John Scott, who argues that in a primordial
atmosphere short on ozone, the ultraviolet radiation would strip electrons away
from water molecules. Such electrons would immediately be surrounded by
four additional water molecules, forming what is termed a hydrated electron.
Before being absorbed into another water molecule, such a hydrated electron
could do a lot of destructive damage to chemical compounds nearby, especially

TO DIG DEEPER

511

those having a net positive charge to which the electron would be attracted.
The essence of Scott’s idea is that the net negative charge of most nucleotides
would offer a protective barrier that would give them a preferential survival
rate in such an environment, their positively charged competitors being wiped
out by the energetic hydrated electrons. Scott presents the details of this case
for a general audience in

Scott, J. “Selection in the Soup.” The Sciences, Nov.-Dee. 1983, pp. 36—42.

For additional details on the role of the hydrated electron, see also the Scott
book noted under General References.

THE CHICKEN'S STORY

For an introductory discussion of Oparin’s coacervates and Fox’s proteinoid
ideas, as well as some personal views expressed by Fox about his critics, see the
Shapiro book cited under General References. Additional material on the pro-
teinoids can be found in

Fox, S. The Emergence of Life. New York: Basic, 1988.

Fox, S. “New Missing Links.” The Sciences, January 1980, pp. 18-21.

Fox, S. “The Proteinoid Theory of the Origin of Life and Competing

Ideas.” American Biology Teacher, 36 (1974), 161-172.

Recent studies indicate that the problems of conducting useful chemical
syntheses in the high-temperature environment of the hydrothermal vents
on the ocean’s floor seem to be insurmountable. For a discussion of the reasons
why, see

Miller, S. L., and J. L. Bada. “Submarine Hot Springs and the Origin of

Life.” Nature, 334 (1988), 609-611.

LIFE: A TWICE-TOLD TALE

It has been argued that the transition from prokaryotic cells to eukaryotic was
the biggest single advance in the whole course of evolution. The current theory
is that it occurred by prokaryotes gobbling up bacteria having useful proper-
ties; so useful, in fact, that the prokaryotes decided not to let them go. Lynn
Margulis has offered virtually incontrovertible evidence that not only did the
mitochondria arise in this fashion, but also the cellular flagellum and the cen-
triole (the device that separates the chromosomes at cell division). Her view
that the hosts and their invaders evolved into a mutually beneficial symbiotic
relationship leading to the eukaryotic cells is detailed in

Margulis, L. Origin of Eukaryotic Cells. New Haven, CT: Yale University

Press, 1970.

Margulis, L. Symbiosis in Cell Evolution. San Francisco: Freeman, 1981.

Shapiro’s proteins-first scheme is given in greater detail in his book noted
earlier. It’s interesting to note Leslie Orgel’s response to Shapiro’s idea. Orgel
commented that he wasn’t too enthusiastic about speculations that didn’t carry
some good experimental evidence along with them — the typical response of ex-
perimentalists everywhere to the unbridled enthusiasms of theoreticians.
Shapiro concedes the point, however, and then goes on to suggest several lines
of experimental attack on the question of whether proteins could, in principle,
come first.

512

TO DIG DEEPER

The second genetic code discussed in the text is written into the structure of
the enzymes that couple the transfer RNA with its corresponding amino acid.
These enzymes, or synthetases, are the real translators between the language of
the proteins and the language of the nucleotides. Recent work suggests that
this code may be much older and more deterministic than the classic genetic
code considered in the text, and it may be less redundant as well. The original
results indicating the possible presence of such a second code are in
Hou, Y.-M., and P. Schimmel. “A Simple Structural Feature Is a Major
Determinant of the Identity of a Transfer RNA.” Nature, 333 (1988),
140-145.

For a less technical summary of the work and its possible implications, see
de Duve, C. “The Second Genetic Code.” Nature, 333 (1988), 117.

“DNA Loses Its Monopoly on Genetic Code.” New Scientist, May 19, 1988,
p. 34.

The short version of Dyson’s “toy model” for the emergence of a system of
metabolizers is presented in his book
Dyson, F. Origins of Life. Cambridge: Cambridge University Press, 1985.

If your interests in the origins question are of the one-nighter variety, this
little masterpiece is the book for you. It offers, in my view, the best possible
introduction to the overall origin-of-life problem, in many ways serving the
same purpose as Schrodinger’s classic What Is Lifef in presenting a modern
physicist’s view of life. The main difference is that Schrodinger centered his
attention upon the process of replication, while Dyson focuses on metabolism.
It’s interesting to note that Schrodinger’s volume served to direct attention to
problems that soon led to the major breakthroughs underpinning much of mod-
ern molecular biology. Perhaps the experimental gaps noted by Dyson will act
in the same manner to stimulate a renaissance in the area of cellular metabo-
lism rather than replication. For a more technical account of Dyson’s ideas, see
Dyson, F. “A Model for the Origin of Life.” Journal of Molecular Evolution,
18 (1982), 344-350.

ASHES TO ASHES, LIFE FROM DUST
The initial suggestion that life might have been based upon silicon in the form
of clays rather than carbon appears to have come from J. D. Bernal, although
he gave them only the secondary role of helping to gather the chemicals needed
to synthesize carbon-based proteins and/or nucleotides.

For an introductory and highly entertaining presentation of Cairns-Smith’s
“seven clues to the origin of life,” see his scientific detective story:
Cairns-Smith, A. G. Seven Clues to the Origin of Life. Cambridge: Cambridge
University Press, 1985.

The technical arguments supporting the conclusions drawn in the above vol-
ume are given in

Cairns-Smith, A. G. Genetic Takeover. Cambridge: Cambridge University
Press, 1982.

TO DIG DEEPER

513

IT CAME FROM OUTER SPACE

It seems that scientists are never happy about the way they’re portrayed by
other scientists, especially in books for a general readership. By all accounts
Crick wasn’t too pleased with Watson’s description of him in The Double
Helix, ostensibly because he didn’t like the idea of personal publicity. As he
tells it, however, he later changed his mind about the book ( including dropping
his idea of a libel suit) because he thought it did a better job than he’d an-
ticipated in showing the general reader how a certain type of scientific research
is done. Later, both Watson and Crick came under fire from Erwin Chargaff of
Columbia, one of the pioneers of molecular biology, who dismissed them with
the withering remark, “In our day that such pygmies throw such giant shad-
ows only shows how late in the day it has become.” There’s just nothing like
public visibility and a Nobel Prize to incite your colleagues’ ire, especially if
you’re young, brash, and seemingly lucky. But as one of my teachers once said,
“I’d rather be lucky than good.”

In explaining the directed panspermia theory, Crick claims that the benevo-
lent aliens would probably send yeast or bacteria as the initial seeds of life, since
these organisms can survive very harsh environments and can live without oxy-
gen. The book was later criticized by the paleontologist Nils Eldredge (of
“punctuated evolution” fame) as being an attack on religion. Crick later argued
that he had nothing against religion, only against beliefs that he feels don’t
correspond to the facts, e.g., antiscientific views, dogmatic fundamentalist
views, irrational views. To see for yourself what he had in mind, “the source” is
Crick, F. Life Itself. New York: Simon and Schuster, 1981.

The popular books outlining the wild visions of Hoyle and Wickramasinghe are
Hoyle, F., and N. C. Wickramasinghe. Diseases from Space. New York:
Harper and Row, 1979.

Hoyle, F., and N. C. Wickramasinghe. Lifecloud. New York: Harper and
Row, 1978.

In all fairness to the H&W comet theory, there is some real scientific evidence
showing that the fundamental idea ( not that proposed by H&W) may be
sound. For a discussion of what needs to be done to settle the matter, see
Bada, J., M. Zhao, and N. Lee. “Did Extraterrestrial Impactors Supply
the Organics Necessary for the Origin of Terrestrial Life? Amino Acid
Evidence in Cretaceous-Tertiary Boundary Sediment.” Origins of Life, 16
(1985-86), 185.

Hobbs, R., and J. Hollis. “Probing the Presently Tenuous Links Between
Comets and the Origin of Life.” Origins of Life, 12 (1982), 125-132.

A very readable and fairly devastating critique of the technical basis of the
H&W Version I theory is given in the Shapiro book noted above. Version II
has not been the object of any kind of scientific critique for obvious reasons.

AND COD CREATED... FROM FISH TO GISH
Like many popular legends, often perpetuated by the filmmakers, the story of
the Scopes Trial as portrayed both in the play and in the movie Inherit the

514

TO DIG DEEPER

Wind is at considerable variance with what actually happened. Contrary to
popular belief, Scopes was not a persecuted biology teacher, but rather a phys-
ical education instructor who was substituting for the regular instructor on
the day in question. More important, Scopes was an enthusiastic participant in
the incident, which had been cooked up by local power brokers as a means of
getting the town of Dayton on the map, as well as to test the constitutionality
of the law in the courtroom. For a fuller account of the real facts surrounding
this bit of Americana, see

Gould, S. J. “A Visit to Dayton,” in Hen’s Teeth and Hone’s Toes, Chap. 20.

New York: Norton, 1983.

The creationist controversy has been extensively treated in so many places
that it’s impossible to do anything other than offer a brief sampler here. For
the position of the creationists themselves, a basic source is

Morris, H. Scientific Creationism. San Diego: CLP Publishers, 1974.

Scientific arguments against the idea of creationism are presented in the fol-
lowing works, of which the next-to-last is especially recommended for its complete
text of Judge Overton’s opinion in the Arkansas case :

But Is It Science f, M. Ruse, ed. Buffalo, NY: Prometheus, 1988.

Gurin, J. “The Creationist Revival.” The Sciences, April 1981, pp. 16-19.

Jukes, T. “Quackery in the Classroom: The Aspirations of the Creationists.”

Journal of Social and Biological Structures, 7 (1984), 193-205.

Kitcher, P. Abusing Science. Cambridge, MA: MIT Press, 1982.

Science and Creationism, A. Montagu, ed. Oxford: Oxford University Press,

1984.

Scientists Confront Creationism, L. Godfrey, ed. New York: Norton, 1983.

While the creationist position is clearly nonscientific, the scientists are not be-
yond reproach in this matter either. Several of the articles in the foregoing com-
pendiums make arguments not so much for science as against the creationists,
on a variety of social, psychological, and political grounds. For example, an ar-
ticle by A. Kehoe in the Godfrey collection begins by giving a nice overview of
the history of the creationists creed, as well as the strategies they have employed
to try to get their ideas institutionalized in the school system. The article then
departs entirely from any sort of “scientific” critique and becomes an emotional
plea for anyone who values the principles upon which the United States is
based to resist the claims of the creationists, since it’s just plain un-American
for any group to have its personal doctrinal beliefs legislated. This article
makes it evident that what is involved here is not a scientific controversy, but
rather a political one. Of course, this has been pretty obvious almost from the
moment the Dayton sheriff put the cuffs on John Scopes, but it’s depressing to
see that the controversy hasn’t really progressed beyond this level, even in the
so-called scientific literature. To my mind, this isn’t a very compelling example
of “scientists” confronting creationism. The Ruse volume is notable for its
emphasis on the philosophical, as opposed to political, aspects of the debate, as
well as the editor’s firsthand account of the Arkansas trial as a participant.

A particularly detailed discussion of the many problems with the classic Pri-
mordial Soup Theory is given in

TO DIG DEEPER

515

Thaxton, 0. , "W. Bradley, and R. Olsen. The Mystery of Life’s Origins. New
York: Philosophical Library, 1984.

In addition to geological, thermodynamic, and chemical evidence against most
of the soup theories, this book also presents an excellent account of the differ-
ence between operations science and origins science. Interestingly, the authors
ultimately end up supporting an off-Earth position on the origins question, but
at least their arguments are cogent and well presented, if somewhat biased
against the conventional wisdom.

THE LOGIC OF LIFE

For more information on von Neumann’s ideas about self-reproducing ma-
chines, see

Essays on Cellular Automata, A. Burks, ed. Urbana, IL: University of Illi-
nois Press, 1970.

von Neumann, J . “The General and Logical Theory of Automata,” in John
von Neumann— Collected Works, Vol. 5, pp. 288-328. New York: Macmillan
1961-63.

Interestingly, the idea of a machine that could make copies of itself and be
“harvested,” much as plants are today, was considered not long after von
Neumann’s original work. A popular account of the economic possibilities is
given in

Moore, E. F. “Artificial Living Plants.” Scientific American, 195 (1956),
118-126.

Moore concludes that such “plants” would have an enormous advantage if we
could solve the design problems, since then we could free agriculture from its
dependence upon the natural characteristics of plants and produce any crop
instead of just those that Nature happens to supply. He ends by noting that he
thinks creation of such an artificial plant would be more easily attainable than
human flight to another planet!

There is by now an extensive literature on the Life game detailing the enor-
mous complexity that can emerge from the very simple rules defining what
happens at each cell. A good introductory presentation, complete with com-
puter programs, is

Poundstone, W. The Recursive Universe. New York: Morrow, 1985.

For those interested the details of Conway’s proof of a self-reproducing Life
pattern, perhaps the most accessible account is that in
Berlekamp, E., J . Conway, and R. Guy. Winning Ways for Your Mathematical
Plays. Volume II. London: Academic Press, 1982.

A natural extension of Conway’s version of Life is to consider playing it in
three dimensions. So instead of the infinite checkerboard, we use an infinite
“egg crate” in which the cells are cubes instead of squares. In many ways this
is a more appropriate version of the game for studying real life, which unfolds
in our three-dimensional space rather than in Conway’s planar world. Interest-
ingly enough, this idea had been suggested as early as 1976 by science fiction

516

TO DIG DEEPER

writer Piers Anthony in his book Ox. Recently, Carter Bays of the University
of South Carolina has explored a wide variety of such three-dimensional ver-
sions of Life. A good introductory account of the difficulties and possibilities,
together with further information, is found in
Dewdney, A. K. The Armchair Universe, pp. 149-159. New York: Freeman,
1988.

Life is by no means the simplest or the most complex cellular automaton
imaginable. In fact, studies of the far simpler one-dimensional automata whose
“universe” consists only of cells on an infinite line, rather than a plane, have
shown equally complex behavior. A good summary of what can happen is given
in the collection.

Wolfram, S. Theory and Applications of Cellular Automata Singapore: World
Scientific, 1986.

The paper by Langton outlining how cellular automata could be used to rep-
resent the functional activities of living entities is
Langton, C. “Studying Artificial Life with Cellular Automata.” Physica D,
22D (1986), 120-149.

Further information on the whole circle of artificial life questions, as well as
an account of some fascinating experiments, can be found in
Dawkins, R. The Blind Watchmaker. Essex, UK: Longman, 1986.

Artificial Life, C. Langton, ed. Reading, MA: Addison-Wesley, 1988.

The following articles represent a selection of material outlining the nature of
computer viruses, as well as some of the difficulties they can cause and what
might be done to create “antiviral” remedies:

Denning, P. “Computer Viruses.” American Scientist, 76 (1988), 236-238.
Dewdney, A. K. “A Core War Bestiary of Viruses, Worms and Other
Threats to Computer Memories.” Scientific American, 252 (1985), 14-23.
Reid, B. “Reflections on Some Recent Widespread Computer Break-ins.”
Communications of the Association for Computing Machinery, 30 (1987), 103-
105.

Witten, I. “Computer (In)security: Infiltrating Open Systems.” Abacus, 4
(1987), 7-25.

CHAPTER THREE
GENERAL REFERENCES

The bible of sociobiology, whose publication sparked off the furor over how we
act, is

Wilson, E. O. Sociobiology: The New Synthesis. Cambridge, MA: Harvard
University Press, 1975.

A good textbook discussion of the principles of sociobiology by one of its fore-
most advocates is

Barash, D. Sociobiology and Behavior. New York: Elsevier, 1977.

TO DIG DEEPER

517

Three books that are must reading for anyone seriously interested in pursuing
the many and varied threads composing the sociobiology “debate” are

Kitcher, P. Vaulting Ambition. Cambridge, MA: MIT Press, 1985.

Ruse, M. Sociobiology: Sense or Nonsense f Dordrecht, Netherlands: Reidel,

1979.

The Sociobiology Debate, A. Caplan, ed. New York: Harper and Row, 1978.

The Caplan volume is a compendium of most of the important papers by the
warring factions in the debate, including Hamilton’s original articles on inclu-
sive fitness and kin selection, the Boston Group’s notorious letter to The New
York Review of Books, and Wilson’s extended response in BioScience, as well as
much, much more. These papers are indispensable reading if you want a clear
view of what created the debate and why it has taken the form that it has. The
book by Michael Ruse is an excellent nonpartisan account by a philosopher of
science assessing the pros and cons of the debate circa 1977. For reasons that
are hard to fathom, Ruse has been labeled a sociobiologist by later commenta-
tors, especially those of the “anti” camp, probably on the grounds that “if
you’re not with us, then you’re against us.” In any case, I find his account to
be a quite impartial, illuminating, thoughtful, and well-written discussion of
all sides of the issue, both scientific and philosophical. Finally, there is the
book by Kitcher. This is another attempt by a philosopher of science to deal
with the whole sociobiology business from the perspective of ten years after-
ward. Some reviews have labeled the book a definitive treatment of the topic,
one that will sound the death knell for sociobiology and close out the debate
.once and for all. Reading these sorts of kudos, I wanted to be enthusiastic
when I first picked up the book, but my expectation of reading an objective
assessment of the facts and theories was dealt a blow when I looked at the
publisher’s dust-jacket blurb and found glowing testimonials by none other
than Richard Lewontin and Stephen Jay Gould — hardly uninvolved or de-
tached observers of the sociobiological scene. After digesting the material, my
view is that a death blow to sociobiology this book is not. Frankly I think
Kitcher, unlike Ruse, has failed to maintain an appropriate arm’s-length dis-
tance from his topic — -a dangerous oversight in a philosophical assessment.
Nonetheless, if you can overlook the author’s somewhat pompous literary style,
there’s a lot of valuable material here and a number of arguments that must be
given serious consideration.

NATURE/NURTURE: SENSE OR NONSENSE?

An excellent description of Milgram’s experimental setup and results is found
in the book

Koestler, A. Janus. New York: Random House, 1978.

Koestler notes the important variation of the experiment in which Milgram
allowed the subjects to inflict any level of shock they wished as a punishment
for a wrong answer, rather than being compelled to use a level determined by
the Leader. In this case, 38 out of the 40 subjects refused to go beyond a level
of 150 volts, the level at which the pupil made his first loud cry, with the
average shock administered a measly 54 volts. Milgram’s own account of these
experiments can be found in his 1974 book Obedience to Authority.

518

TO DIG DEEPER

NEO-NEO-DARWINISM AND SOCIOBIOLOGY
Library shelves groan under the weight of books expounding the Darwinian
and neo-Darwinian theories of evolution, so let me content myself here with the
following short, well-written and easily accessible sources:

Arthur, W. Theories of Life. London: Penguin, 1987.

Ayala, F. “The Mechanisms of Evolution.” Scientific American, 239 (Septem-
ber 1978), 56-69.

Smith, J. Maynard. Problems of Biology. Oxford: Oxford University Press,
1986.

The emergence of sociobiology as an interdisciplinary amalgam of ethology,
population ecology and evolutionary genetics is traced in

Barlow, G. “The Development of Sociobiology: A Biologist’s Perspective,”
in Sociobiology: Beyond Nature/Nurture 1, G. Barlow and J. Silverberg, eds.,
pp. 3-24. Boulder, CO: Westview Press, 1980.

The distinction between the Central Dogma of Molecular Biology and what
I’ve termed here the Central Dogma of Social and Behavioral Biology can be
made more explicit by the following diagram:

GENETIC INHERITANCE

transcription
DNA ^ s~\

j ■+-§)—

RNA

reverse

transcription

translation
► Protein

^ Epigenetic
rules

G CULTURAL INHERITANCE

transcription

antic Learned translation ^ Artifacts,

rork ◄ behavior lifeways

reverse

transcription

The dogmas of genetic and cultural inheritance

Here the prohibition against information flow from the proteins to the geno-
type is indicated by the x in reverse transcription for genetic inheritance. On
the other hand, while DNA can replicate itself, its cultural equivalent, the se-
mantic network, cannot. This diagram also shows the connection between the
epigenetic rules of Wilson and Lumsden and the processes of genetic and cul-
tural inheritance.

ANIMAL ANTICS

A thorough discussion of game theory in the evolutionary context by the mas-
ter himself is given in

Smith, J . Maynard. Evolution and the Theory of Games. Cambridge: Cam-
bridge University Press, 1982.

TO DIG DEEPER

519

An introductory, textbook-level account is found in Chapter Six of

Casti, J. Alternate Realities: Mathematical Models of Nature and Man. New
York: Wiley, 1989.

Pull details of the experiments by Riechert on ESS for grassland spiders are
reported in

Riechert, S. “Spider Fights as a Test of Evolutionary Game Theory.” Amer-
ican Scientist, 74 (1986), 604-610.

The discussion of the parental investment question, together with a much
more detailed (and entertaining) treatment of the “arms race” between males
and females, is given in

Dawkins, R. The Selfish Gene. Oxford: Oxford University Press, 1976.

A nice account of the sex determination procedure in Hymenoptera, as well
as an easily readable discussion of associated matters like altruism and inclu-
sive fitness is found in

Smith, J. Maynard. “The Evolution of Behavior.” Scientific American, 239
(September 1978), 176-192.

For the source papers in which Hamilton introduced the notion of inclusive
fitness, see the Caplan book noted earlier.

For a discussion of why there might be no more to be learned about people
from observing animals than by reading Aesop’s fables, see the article
Simon, M. “Sociobiology: The Aesop’s Fables of Science.” The Sciences, 18
(1978), 18-21.

THE STRANGE CASE OF ALTRUISM
The original paper by Trivers that outlined the case for reciprocal altruism is
Trivers, R. “The Evolution of Reciprocal Altruism.” Quarterly Review of Bi-
ology, 46 (1971), 35-39, 45-17

THE GENETIC IMPERATIVE

Like Darwin, who devoted only a few words in his epic works to the special
problems of human evolution, in Sociobiology Wilson addresses the matter of
human sociobiology only in the book’s last chapter, and in a purely speculative
mode. However, also like Darwin, Wilson had clearly been thinking long and
hard about the implications of his work for Homo sapiens, as shown in his
full-length treatment of the matter in

Wilson, E. O. On Human Nature. Cambridge, MA: Harvard University
Press, 1978.

For Wilson’s personal statement about many of the ideas put forth in this
work, see the interview

“Genetic Destiny,” Omni, 1978.

The coevolutionary circuit of Lumsden and Wilson is completely described
in all its painstaking mathematical and sociobiological detail in

Lumsden, C., and E. O. Wilson. Genes, Mind, and Culture. Cambridge, MA:
Harvard University Press, 1981.

520

TO DIG DEEPER

See also the updated treatment given in the paper
Lumsden, C., and E. O. Wilson. “The Relation Between Biological and
Cultural Evolution.” Journal of Social and Biological Structures, 8 (1985),
343-359.

Stephen Jay Gould has eloquently put forward a case for the flexibility of
the human brain as the principal reason why it’s not necessary to invoke a
genetic explanation for human behavior. A popular account of his argument is
found in

Gould, S. J. “Biological Potentiality vs. Biological Determinism,” in Ever
Since Darwin, pp. 251-259. New York: Norton, 1977.

GETTING INTO HER GENES

As a good example of the kind of direct attacks in the literature claiming that
sociobiology is sexist, see

Alpher, J., J. Beckwith, and L. Miller. “Sociobiology Is a Political Issue,”
in The Sociobiology Debate, cited above.

Excellent popular accounts expounding Wilson’s views of the biological ori-
gin of religion and morals are

Masters, R. “Sociobiology: Science or Myth?” Journal of Social and Biological
Structures, 2 (1979), 245-252.

Wilson, E. O. “Human Decency Is Animal.” New York Times Magazine, Oc-
tober 12, 1975.

Somewhat more detailed discussions are given in the books
Flanagan, O. The Science of the Mind. Cambridge, MA: MIT Press, 1984.
Schwartz, B. The Battle for Human Nature. New York: Norton, 1986.
von Schilcher, F., and N. Tennant. Philosophy, Evolution and Human Nature.
London: Routledge and Kegan Paul, 1984.

Each of these books is quite remarkable in its own way, giving a critical pic-
ture of sociobiology from a particular vantage point. Flanagan speaks as a
philosopher of science, emphasizing the critical arguments against Wilson’s
claims for the origin of morality and normative principles out of biological
necessity. Schwartz focuses his attention on the doctrine of self-interest as it
arises in Adam Smith’s economics, the evolutionary biology of Darwin, and
Skinner’s behavioristic views in psychology. Within this setting, sociobiology
is treated primarily as an attempt to show that the behavior of animals (in-
cluding humans) serves reproductive fitness, by applying the notion of eco-
nomic self-interest to social behavior. Finally, the von Schilcher and Tennant
book is a critical analysis of modern evolutionary theory, assessing its philo-
sophical consequences in relation to morality, knowledge, consciousness, and
language with special attention to problems of cultural evolution. Taken to-
gether, these volumes provide extremely good coverage of sociobiology as seen
from the philosopher’s point of view.

CANT VS. KANT

For an easily digestible recounting of the story of social Darwinism in Amer-
ica, see

Morris, R. Evolution and Human Nature. New York: Putnam, 1983.

TO DIG DEEPER

521

A good journalistic report from the sociobiology front written at the time
the accusations were flying hot and heavy is
Wade, N. “Sociobiology: Troubled Birth for New Discipline.” Science, 191
(March 19, 1976), 1151-1155.

Apparently, the first that Wilson knew of the Boston Group’s attack was when
he received a phone call from science journalist Boyce Rensberger asking for
his reaction. Wilson, of course, was dumbfounded at the fact that The New
York Times had a copy of an attack that had been prepared by colleagues whom
he regarded as friends and who, moreover, occupied offices within a few hun-
dred meters of his own. For an excellent discussion of the circumstances sur-
rounding the infamous New York Review of Books letter, as well as a compact
summary of the claims and counterclaims, see
Currier, R. “Sociobiology: The New Heresy.” Human Behavior, Nov. 1976,
16-22.

Another good source for what the debate between Wilson and his colleagues is
all about is

Ruse, M. “Sociobiology: Sound Science or Muddled Metaphysics?” Proceed-
ings of the 1976 Philosophy of Science Association Meeting, F. Suppe and P.
Asquith, eds., pp. 48-73. East Lansing, MI: Philosophy of Science Associa-
tion, 1977,

The original letter of the Boston Group to The New York Review of Books is
reprinted in the Caplan compendium noted earlier. A more extensive version of
this critique can be found in

Allen, E., et al. “Sociobiology: Another Biological Determinism.” BioScience,
26 (1976), 182-186.

Wilson’s responses to the two attacks are reported in
Wilson, E. O. “Academic Vigilantism and the Political Significance of Soci-
obiology.” BioScience, 26 (1976), 183-190.

Wilson, E. O. “Letter to the Editor,” New York Review of Books, 22 (1975),
No. 20, 60-61.

On the other side of the Atlantic, Lewontin’s British comrades in arms were
also not hesitant to chip in with their own two cents’ (or pence) worth of criti-
cism of Wilson, as well as of their countryman Richard Dawkins. A couple of
representative samples are

Midgley, M. “Gene-Juggling.” Philosophy, 54 (October 1979).

Rose, S. “Pre-Copernican Sociobiology?” New Scientist, October 5, 1978,
45-46.

In his inimitable style, Dawkins replies to an earlier claim of Rose’s that his
work fosters racism and neo-Nazi ideals in
Dawkins, R. “Selfish Genes in Race or Politics.” Nature, 289 (1981), 528.

See also his extended response to the hostility of Midgley’s claims in
Dawkins, R. “In Defence of Selfish Genes.” Philosophy, 56 (1981), 556-573.

For a discussion of the political views of Lewontin and their intertwining
with his biological work, see

522

TO DIG DEEPER

Lumsden, C., and E. 0. Wilson. “Genes, Mind, and Ideology.” The Sciences,
21 (November 1981), 6-8.

Lewontin’s remark about taking his job as a political activity can be found in
the Chronicle of Higher Education, October 23, 1973.

A full-scale attack on sociobiology from a biological as well as political
standpoint is contained in

Biology as a Social Weapon, Science for the People Collective, eds. Minneapo-
lis, MN: Burgess, 1977.

Lewontin, R., S. Rose, and L. Kamin. Not in Our Genes. New York: Pan-
theon, 1984.

Interestingly enough, prior to the offensive launched by the Boston Group, the
professional reviews of Wilson’s book Sociobiology had been quite favorable.
Good examples are found in the following collection:

“Multiple Reviews of Wilson’s Sociobiology.” Animal Behavior, 24 (1976),
698-718.

In fact, of the fourteen reviewers contributing to the above collection, only one
definitely comes down on the negative side of the ledger. While speaking of
reviews, of special interest is the review by Elliott White of Kitcher’s inflam-
matory book cited under General References above. In this discussion, White
argues convincingly against the egalitarian basis of many of the Boston
Group’s most bitter complaints against Wilson. For the full review, see

White, E. “Review of Kitcher, P., Vaulting Ambition: Sociobiology and the
Quest for Human Nature.” Journal of Social and Biological Structures, 11
(1988), 283-286.

SO-SO BIOLOGY

Sahlins’s criticism of the practical aspects of kin selection are contained in his
scathing critique of sociobiology:

Sahlins, M. The Use and Abuse of Biology: An Anthropological Critique of Socio-
biology. Ann Arbor, MI: University of Michigan Press, 1976.

Another critique of the idea of biological determinism worth noting is

Thompson, J. “Human Nature and Social Explanation,” in Against Biologi-
cal Determinism, S. Rose, ed., pp. 30—49. London: Allison and Busby, 1982.

Dawkins’s idea of a hereditary agent playing the role in culture that genes
do in biology has been put forward a number of times. In addition to the meme
presented in Dawkins’s book The Selfish Gene, and the culturgen of Lumsden
and Wilson, the same concept has been discussed under the rubric of a socio-
gene in

Swanson, C. Ever-Expanding Horizons. Amherst, MA: University Massachu-
setts Press, 1983.

The idea of using biological evolutionary concepts to try to model the process
of cultural change has also been pursued by many authors, some with a ven-
geance. Two relatively recent treatments showing the kind of mathematical
level to which the idea has risen are

TO DIG DEEPER

523

Boyd, R., and P. Richerson. Culture and the Evolutionary Process. Chicago:
University of Chicago Press, 1985.

Cavalli-Sforza, L., and M. Feldman. Cultural Transmission and Evolution: A
Quantitative Approach. Princeton, NJ: Princeton University Press, 1981.

CONFLICTING RATIONALITIES AND
THE DILEMMA OF COOPERATION
By far the best non-mathematical, introductory account I know of treating
game theory for the social scientist is

Colman, A. Game Theory and Experimental Games. Oxford: Pergamon Press,
1982.

This book is filled with interesting examples of different types of games model-
ing every kind of human strategic interaction from arms races to confronta-
tions over moral philosophy. But if you want the mathematics behind the
results discussed in Colman, you’ll have to go elsewhere. One good place is
Jones, A. J. Game Theory: Mathematical Models of Conflict. Chichester, UK:
Ellis Horwood, 1980.

The Prisoner’s Dilemma has by now been the subject of well over one thou-
sand research articles and numerous book-length accounts. Still one of the
best is

Rapoport, A., and A. Chammah. Prisoner’s Dilemma: A Study in Conflict and
Cooperation. Ann Arbor, MI: University of Michigan Press, 1965.

The fascinating computer tournaments of Axelrod are described in
Axelrod, R. The Evolution of Cooperation. New York: Basic, 1984.

See also the easily accessible popular discussion in
Hofstadter, D. “Computer Tournaments of the Prisoner’s Dilemma,” in
Metamagical Themas, pp. 715-734. New York: Basic, 1985.

In a related work, Peter Corning argues for the idea of egoistic cooperation
as a theory of progressive evolution. Corning notes that in a world of 2 mil-
lion living species, only about ten thousand can be said to be eusocial. He asks
how such islands of cooperation can emerge in a sea of conflict. For his an-
swer see _

Corning, P. The Synergism Hypothesis. New York: McGraw-Hill, 1983.

An introductory account of Axelrod’s more recent work on the Norms Game is
found in

Axelrod, R. “Laws of Life.” The Sciences, 27 (1987), No. 2, 44-51.
BRINGING IN THE VERDICT

During the course of reviewing Melvin Konner’s book The Tangled Wing, an
extended meditation on the biology of human emotions, the noted science
journalist Horace Freeland Judson reviews many of the attacks on socio-
biology, concluding that it has by no means lost the war. His arguments are
given in

Judson, H. F. “An Imperial Presence.” The Sciences, 23 (1983), 20-23.

524

TO DIG DEEPER

CHAPTER FOUR

GENERAL REFERENCES

An encyclopedic source (literally) for information about all aspects of lan-
guage is

The Cambridge Encyclopedia of Language, D. Crystal, ed. Cambridge: Cam-
bridge University Press, 1987.

An exploration of the thesis that language is really the interplay between sys-
tems of grammar and human behavior is carried out in the following very
readable, almost popular, volume:

Farb, P. Word Play: What Happens When People Talk. New York: Knopf,
1974.

The interconnections of information theory, languages, and codes like DNA are
presented in a form suitable for popular consumption in
Campbell, J. Grammatical Man. New York: Simon and Schuster, 1982.

A standard textbook account of language in its many manifestations is
Fromkin, V., and R. Rodman. An Introduction to Language, 3rd Edition. New
York: Holt, Rinehart and Winston, 1983.

For a Trivial Pursuit-type miscellany of fascinating facts about the peculiari-
ties of the world’s languages, such as the fact that German was almost adopted
as the official language of the United States, or that the complete form of the
Spanish insult itu madre! consists of five syllables that are often just whistled
or beeped out on the horn of a car, see
Berlitz, C. Native Tongues. New York: Grosset and Dunlap, 1982.

A detailed account of all the major schools of linguistic thought from de Saus-
sure to the modern London school is provided in
Sampson, G. Schools of Linguistics. Stanford, CA: Stanford University Press,
1980.

An ever-increasing amount of evidence is coming to light suggesting that
human language origins are biologically based in evolutionary changes in our
vocal mechanisms, along with corresponding changes in neural control circuits
in the brain. One of the prime exponents of this view is Philip Lieberman of
Brown University, who gives a nontechnical introduction to his ideas in
Lieberman, P. “Voice in the Wilderness.” The Sciences, 28, No. 4 (1988),
23-29.

For more technical accounts of the same type, but applied to the even broader
issues of general intelligence, see

Intelligence and Evolutionary Biology, H. and I. Jerison, eds. Berlin:
Springer, 1988.

DUMB DOCS AND CLEVER HANS

Interspecies communication seems to hold a continuing fascination for humans
of all ages, an instinctual urge that can be seen by our predilection for keeping

TO DIG DEEPER

525

house pets. Several works detailing the current state of this ongoing effort to
talk with the animals are

Animal Intelligence: Insights into the Animal Mind, R. Hoage and L. Gold-
man, eds. Washington, D.C.: Smithsonian Press, 1986.

The Clever Hans Phenomenon: Communication with Horses, Whales, Apes, and
People, T. Sebeok and R. Rosenthal, eds., Annals of the New York Academy
of Sciences, Vol. 364. New York: New York Academy of Sciences, 1981.
Crail, T. Apetalk & Whalespeak: The Quest for Interspecies Communication.
Chicago: Contemporary Books, 1983.

Griffin, D. Animal Thinking. Cambridge, MA: Harvard University Press,
1984.

Griffin, D. The Question of Animal Awareness, Revised Edition. Los Altos, CA:
Kaufman, 1981.

Wade, N. “Does Man Alone Have Language? Apes Reply in Riddles, and a
Horse Says Neigh.” Science, 208 (June 20, 1980), 1349-1351.

The Hoage and Goldman volume contains the papers presented at a 1983 sym-
posium addressing the issues of animal cognition. It represents an excellent
survey of the entire field by the practitioners themselves. The Clever Hans book
and the Wade article zero in not only on the question of animal communica-
tion, but also on the equally important problem of investigator deception. How
can we really separate the effects of real animal communication from the kinds
of cues given by their masters, wittingly or not? Crail’s book is a popular in-
troduction to the entire program of research on animal communication, rang-
ing from the Gardners’ work with chimpanzees to Lilly’s efforts to
communicate with the dolphins. The two books by Griffin discuss his lifelong
efforts to try to understand the cognitive processes of animals and the question
of whether or not it makes sense to speak of animal consciousness. Taken to-
gether, these items cover just about everything that an interested reader would
need to know to get to the forefront of current research on this eternally tanta-
lizing topic.

VERBAL BOTANY AND UNIVERSAL GRAMMAR
A quick overview of the development of linguistics as a science is given for the
general reader in

Gardner, H. The Mind’s Hew Science. New York: Basic, 1985.

This volume also serves as the best possible nontechnical introduction to the
entire area now covered by the umbrella term cognitive science. For a more de-
tailed look at linguistics per se, see the Sampson book cited earlier.

A good discussion of the entire problem of language acquisition, albeit from
a decidedly Chomskian point of view, can be found in

Lightfoot, D. The Language Lottery. Cambridge, MA: MIT Press, 1982.

According to linguistic folklore, Chomsky’s original manuscript, “The Logi-
cal Principles of Linguistic Theory,” was prepared during his tenure as a jun-
ior fellow at Harvard. The MIT Press declined to issue the work and, as the
story goes, a representative of the Dutch house Mouton picked up vibrations

526

TO DIG DEEPER

about the book from one of its representatives who was curious about the ex-
cerpted version that Chomsky was then using for his classes at MIT. The rest
is history. The actual citation for this pathbreaking work is
Chomsky, N. Syntactic Structures. The Hague: Mouton, 1957.

THE NOAM OF CAMBRIDGE

By now, Chomsky’s ideas about language, mind, politics, and life have been
chronicled in so many places and in so many different ways that there’s liter-
ally an account for every intellectual taste and purse. Probably much of the
reason for this widespread interest in his ideas is due to his role as one of the
most vocal opponents of the U.S. policy in Vietnam. In fact, one reporter for
The New York Times was surprised to find that Chomsky was a famous linguist
and that his linguistics had something to do with his public role as a political
figure. Since I personally don’t find that much connection between his linguis-
tics and his politics, I haven’t dwelt upon the latter in this chapter. However,
for those readers wanting more details in this direction, as well as full accounts
of the Chomskian revolution, linguistically speaking, two of the best sources
are the biographies

Leiber, J. Noam Chomsky. Boston: G. K. Hall, 1975.

Lyons, J. Noam Chomsky, Revised Edition. London: Penguin, 1977.

The Lyons book is readily available worldwide in paperback and gives an excel-
lent introductory account of Chomsky’s life and thoughts. However, for those
wanting more than just a surface account of the ideas, but without the back-
ground or interest for attacking a full-scale technical treatment, the book by
Leiber is hard to beat. Too bad it’s so difficult to find. But the search is defi-
nitely worth the effort. For a verbatim account of Chomsky’s views on linguis-
tics, psychology, sociobiology, Piaget, Skinner and much more, see
Gliedman, J. “Interview with Noam Chomsky.” Omni, 1979.

“The Ideas of Chomsky,” in Men of Ideas, B. Magee, ed. Oxford: Oxford
University Press, 1978.

Relatively accessible technical accounts of transformational grammars are
given in the Lightfoot book noted above, as well as in
Smith, N., and D. Wilson. Modem Linguistics: The Results of the Chomsky’s
Revolution. Bloomington, IN: Indiana University Press, 1979.

Perhaps the most readable discussions by Chomsky himself on these matters
are contained in his general lectures:

Chomsky, N. Language and Mind, Enlarged Edition. New York: Harcourt,
Brace, Jovanovich, 1972.

Chomsky, N. Reflections on Language. New York: Pantheon, 1975.

A critical assessment of Chomsky’s theories on linguistics as they stood at
the end of the 1970s is found in the collection
On Noam Chomsky: Critical Essays, 2nd Edition, G. Harman, ed. Amherst,
MA: University of Massachusetts Press, 1982.

Of special interest in this book are the articles by John Searle and Robert
Lees, the first a reprint of Searle’s well-known 1972 article in The New York

TO DIG DEEPER

527

Review of Books, which serves as an eminently readable introduction to the
whole corpus of Chomsky’s thoughts in linguistics. The Lees contribution is
the review of Syntactic Structures in the journal Language that sparked off the
Chomskian revolution. It’s perhaps not without interest to note that Lees, as
well as being a linguist, is also a chemical engineer and was working at the
MIT Research Lab of Electronics at the time of writing this review. Thus, he
was uniquely prepared to understand and appreciate what was at the time the
novel, almost engineering-oriented nature of Chomsky’s approach. As a fur-
ther offering from this very informative volume, let me quote the poem by
John Hollander showing that Chomsky’s famous “colorless green ideas sleep
furiously” may have semantic content, or at least utility, after all:

COILED ALIZARINE

for Noam Chomsky

Curiously deep, the slumber of crimson thoughts:

While breathless, in stodgy viridian,

Colorless green ideas sleep furiously

POSITIVELY REINFORCING

Skinner’s ideas on behavior and mind have by now entered into what one could
almost term the folk wisdom of American popular psychology, having been
explicated in innumerable books and articles. A worthwhile recent account put-
ting Skinner’s behaviorist notions into the context of modern ideas on thought
and mind is given in

Flanagan, O. The Science of the Mind. Cambridge, MA: MIT Press, 1984.

This book, incidentally, is also an excellent reference for the ideas of Piaget
and their relationship to the mainstream of current thinking on minds and
machines.

Chomsky’s notorious review of Skinner’s Verbal Behavior was originally
published in the widely circulated periodical Language, and served as one of the
major stepping-stones for the ascendancy of Chomskian ideas into the domi-
nant position not only in modern linguistics, but in psychology as well. The
original review is

Chomsky, N. “Review of Skinner’s Verbal Behavior.” Language, 35 (1959),
26-58.

Undaunted by the decline of behaviorism as a significant line of thought in
modern psychology, even in retirement Skinner continues to not only preach
the behaviorist gospel, but also to practice what he preaches by living his daily
life in a modern version of his Skinner box. For a journalistic account of Skin-
ner at age eighty-three, see

Goleman, D. “The Behaviorist Box of B. F. Skinner.” International Herald
Tribune, August 28, 1987.

OUT OF THE MOUTHS OF BABES

Piaget is usually counted as one of the founders of the so-called structuralist
school of thinkers, another being the famed French anthropologist Claude

528

TO DIG DEEPER

Levi-Strauss. Interestingly, these two pioneers took diametrically opposed po-
sitions on the role of language in shaping thought processes. Piaget, as we
know, felt that language makes only a small contribution to thought, while
Levi-Strauss was of the opinion that one starts with language, which then
plays a determining role in thought. For an account of both men and their
lives, work, and roles in the development of the structuralist movement, the
following book is hard to beat:

Gardner, H. The Quest for Mind: Piaget, Levi-Strauss, and the Structuralist

Movement, 2nd Edition. Chicago: University of Chicago Press, 1981.

For another good source of critical analysis of Piaget’s role in establishing the
cognitive thrust of modern psychology and its consequent influence on theories
of the mind, see the Flanagan book cited in the preceding section. This book
also provides a good account of the developmental theories of the psychologist
Lawrence Kohlberg regarding the stages of evolution of morals. In Kohlberg’s
view, there is an objective moral “good,” which becomes apparent in a half-
dozen or so stages of development. Basing his theory of moral development
upon Piaget’s stages of cognitive development, Kohlberg claims to be able to
resolve the debate between the Kantians, who cling to an absolute categorical
imperative, and the followers of Mill, who argue for a kind of pleasure-maxi-
mizing utilitarianism. According to Kohlberg’s extension of Piaget’s stages,
the hands-down winner of this particular fight is Kant.

IT'S ALL A QUESTION OF SEMANTICS
The work of Sapir and Whorf contending that one’s view of the world is not
only influenced but actually determined by one’s language is outlined in the
Sampson book noted earlier under General References. For Whorf’s own ac-
count, see the following collection of reprints of his articles:

Language , Thought, and Reality: Selected Writings of Benjamin Whorf, J. B.

Carroll, ed. Cambridge, MA: MIT Press, 1956.

A penetrating discussion of the relevance of Chomsky’s ideas vis-a-vis those
of relativists like Sapir and Whorf within the context of literary analysis is
given in

Steiner, G. “Whorf, Chomsky, and the Student of Literature,” in On Diffi-
culty: Selected Essays. Oxford: Oxford University Press, 1978.

An introduction to the work of Sampson on an evolutionary approach to
linguistics is given in

von Schilcher, F., and N . Tennant. Philosophy, Evolution and Human Nature.

London: Routledge and Kegan Paul, 1984.

For a more extensive discussion, see

Sampson, G. “Linguistic Universals as Evidence for Empiricism.” Journal of

Linguistics, 14 (1978), 129-375.

Sampson, G. Making Sense. Oxford: Oxford University Press, 1980.

For an assessment of some of Sampson’s views in the above volume, see the
following review, which questions Sampson’s ability to deal with the “poverty
of the stimulus” problem:

TO DIG DEEPER

529

Lightfoot, D. “Review of Making Sense,” in Journal of Linguistics, 18 (1982),
426—431.

The Watchmaker Parable underlying Sampson’s evolutionary approach to
the building-up of a hierarchical language structure is presented in
Simon, H. “The Architecture of Complexity,” in The Sciences of the Artificial,
2nd Edition, Cambridge, MA: MIT Press, 1981.

From a computational point of view, the result of Peters and Ritchie shows
that a Chomskian transformational grammar is capable of computing (in the
formal sense made specific in the artificial intelligence chapter) anything that
can be computed. A strong argument can also be made that to survive, humans
must also be able to compute in some abstract sense. The question is then how
much computing power we really need in order to survive. Since presumably
evolution has endowed us with computing power “from below,” we embody an
amount of computing capability that’s sufficient for our needs, but little more.
Some have claimed, therefore, that it’s unreasonable to suppose that our brains
must necessarily be modeled by the most powerful type of computing machine
that’s theoretically possible. It’s at this point that the Montague grammars,
with their computational limitations to characterizing only context-sensitive
languages, begin to look interesting. For further technical details on the struc-
ture of such grammars, see

Montague, R. Formal Philosophy. New Haven, CT: Yale University Press,
1974.

For a summary of recent work building upon the foundations laid by Mon-
tague, see

Gazdar, G. “Generative Grammar,” in New Horizons in Linguistics, Vol. 2, J.
Lyons et al, eds., pp. 122-151. London: Penguin, 1987.

SHOOT-OUT AT THE ROYAUMONT CORRAL
The definitive account of the goings-on at Royaumont is given in
Language and Learning: The Debate Between Jean Piaget and Noam Chomsky,
M. Piattelli-Palmarini, ed. Cambridge, MA: Harvard University Press,
1980.

This volume presents not only the salvos fired by both of the principals, but
also extensive rumblings from the “chorus,” as well as detailed postmortems
by other commentators on the cognitive science scene. It’s interesting to note
that the debate resulted in a living case study of Piaget’s ideas of accommoda-
tion and assimilation, since Chomsky seemed to insist that others accommo-
date their own views to his own, while Piaget held open the possibility of wid-
ening his own views to assimilate the Chomskian criticisms into his system.
Probably the best that can be said about the outcome is that the two views
are complementary in much the same way that waves and particles are com-
plementary in quantum theory. A popularized account of the debate can be
found in

Gardner, H. “Encounter at Royaumont.” Psychology Today, July 1979, pp.
14-16.

530

TO DIG DEEPER

Illuminating and compact introductions to Chomsky’s current thinking on
the mind can be found in his recent semipopular accounts based on lecture
series in San Diego and Managua:

Chomsky, N. Language and Problems of Knowledge: The Managua Lectures.
Cambridge, MA: MIT Press, 1988.

Chomsky, N. Modular Approaches to the Study of the Mind. San Diego, CA:
San Diego State University Press, 1984.

In the San Diego lectures, Chomsky gives a particularly concise summary of
the problems surrounding mental representations, assuming they exist. Ac-
cording to his account, they can be divided into three categories:

• The Syntax Problem: Of what kinds of elements are the representations com-
posed and how are they put together!

• The System Problem: How are the various mental modules organized and in-
terconnected!

• The Rule Problem: Can we characterize mental representations in terms of a
system of rules that determines their properties!

RULES AND REPRESENTATIONS

The claim that human cognitive faculties can be described by rules acting on
mental representations is the very essence of the machine metaphor that under-
pins the hopes of the artificial intelligentsia in particular, and the cognitive
scientists in general. For a nice textbook introduction to cognitive science, see
Stillings, N., et al. Cognitive Science: An Introduction. Cambridge, MA: MIT
Press, 1987.

An extensive account of Chomsky’s ideas on the question of rules and mental
representations is presented in his book Rules and Representations (New York:
Columbia University Press, 1980). His major points are excerpted, together
with extensive peer commentary, in

Chomsky, N. “Rules and Representations.” Behavioral and Brain Sciences, 3
(1980), 1-61.

The system-theoretic perspective showing the essential equivalence of exter-
nal and internal rules, at least from a mathematical point of view, is developed
in detail in

Casti, J. “Behaviorism to Cognition: A System-Theoretic Inquiry into
Brains, Minds, and Mechanisms,” in Real Brains, Artificial Minds, J. Casti
and A. Karlqvist, eds., pp. 47-75. New York: Elsevier, 1987.

CHAPTER FIVE
GENERAL REFERENCES

For a general overview of current AI principles and practice, the following
books are particularly good, giving an easily accessible account of many of the
ideas and actors on today’s AI scene:

Johnson, G. Machinery of the Mind. New York: Times Books, 1986.
Waldrop, M. Man-Made Minds. New York: Walker, 1987.

TO DIG DEEPER

531

For an easily understood introduction to some of the technical ideas that I’ve
only touched upon, see

Aleksander, I., and P. Burnett. Thinking Machines. Oxford: Oxford Univer-
sity Press, 1987.

Haugeland, J. Artificial Intelligence: The Very Idea. Cambridge, MA: MIT
Press, 1985.

Much of the early history of AI up to the mid-seventies, as well as in-depth
interviews and portraits of many of the players in our game, such as Simon,
Newell, Dreyfus, and Feigenbaum, is given in the work
McCorduck, P. Machines Who Think. San Francisco: Freeman, 1979.

In 1983 the New York Academy of Sciences sponsored a meeting devoted to
all aspects of the scientific, intellectual, and social impact of the computer.
Part of that workshop was a round-table discussion on the question of what we
have termed strong AI, human. The transcript of that discussion provides a
good background to the entire spectrum of matters considered here. It can be
found in the volume

Computer Culture, H. Pagels, ed., Annals of the New York Academy of
Sciences, Vol. 426. New York: New York Academy of Sciences, 1984.

The systems interface of AI, neuroscience, and cognitive psychology, to-
gether with an exposition of some of the top-down and bottom-up conflicts, is
explored in

Boden, M. Computer Models of the Mind: Computational Approaches in Theoreti-
cal Psychology. Cambridge: Cambridge University Press, 1988.

Mindwaves, C. Blakemore and S. Greenfield, eds. Oxford: Blackwell, 1987.
Real Brains, Artificial Minds, J. Casti and A. Karlqvist, eds. New York: El-
sevier, 1987.

The theme of thinking machines and their possible technological, social and
psychological implications for man has long been a staple of the science fiction
community. Some of my favorites in this line are

Hogan, J. P. Two Faces of Tomorrow. New York: Ballantine, 1979.

Jones, D. F. Colossus. New York: Berkeley, 1976.

Ryan, T. J. The Adolescence of Pi. New York: Macmillan, 1977.

Each of these books deals with the general theme of a cognitive computer run
amok, threatening human supremacy, and finally yielding its usurped control
back to its human masters. It’s probably stories like these that give the Wei-
zenbaums of the world nightmares, but for the rest of us they offer a sugar-
coated lesson in how thinking machines might actually come about, and the
kinds of behavior they might display.

THE TURING TEST AND THE CHINESE ROOM
The Imitation Game was first suggested by Alan Turing in the fundamental
paper

Turing, A. “Computing Machinery and Intelligence.” Mind, 59 (1950).

This paper has since been reprinted in many places, perhaps the most easily
accessible being

532

TO DIG DEEPER

Hofstadter, D., and D. Dennett. The Mind’s I. New York: Basic, 1981.

This volume is highly recommended as a treasure trove of additional original
material, together with extensive editorial commentary, on the entire spectrum
of issues pertaining to minds, brains, machines, souls, and self.

Searle’s original paper in which he presents the Chinese Boom thought ex-
periment is

Searle, J. “Minds, Brains, and Programs.” Behavioral and Brain Science, 3

(1980), 417-424.

This already classic paper has also been reprinted a number of times, including
an appearance in the Hofstadter and Dennett volume just cited. However, I
strongly recommend the original reference as it also contains extensive peer
commentary by twenty-seven of the most prominent workers in the field, as
well as Searle’s rejoinder to their remarks.

Alan Turing was truly one of the unsung heros of the Second World War,
his breaking of the German command code ranking with the development of
the atomic bomb as a pivotal factor in the war’s outcome. However, in contrast
to von Neumann, Oppenheimer, Teller, & Co., Turing and his work both faded
into a totally undeserved obscurity following the war, with even his position in
academic circles being relatively anonymous. It is only in the last decade or so
that Turing’s real genius has been given public recognition, much of it attrib-
utable to the outstanding biography:

Hodges, A. Alan Turing: The Enigma. New York: Simon and Schuster, 1983.

This work tracing Turing’s life and career, together with his tragic suicide,
has recently been produced as the play Breaking the Code, which has had a
successful run on the London and New York stages, further exposing to the
general public Turing’s fundamental contributions both to science and to his
country.

The work begun in the late 1940s examining the interface between brains
and machines represents the germ of the idea that is now flourishing under the
rubric “cognitive science.” An excellent account for the general reader of the
history, objectives, and current programs in this field is

Gardner, H. The Mind’s New Science: A History of the Cognitive Revolution.

New York: Basic, 1985.

FORMAL SYSTEMS, MACHINES, AND TRUTHS
A wonderful introduction to the charms and wiles of formal systems as well as
much, much more is the tour de force

Hofstadter, D. Godel, Escher, Bach: An Eternal Oolden Braid. New York:

Basic, 1979.

In this Pulitzer Prize-winning masterpiece, Hofstadter introduces what
amounts to the manifesto of the bottom-up school of AI by means of a series of
Lewis Carroll-like dialogues, thought experiments, and philosophical specula-
tions elucidating the intricacies of formal systems, Turing machines, Godel’s

TO DIG DEEPER

533

theorems, Zeno’s paradoxes, the Turing-Church Thesis, the theory of evolu-
tion, self-reference, and a whole lot more. Hofstadter probably did irreparable
harm to his standing in the mainstream AI community by having the temerity
to write such a call to arms, especially one committing what in academia are
the cardinal sins of being both intelligible and popular with the public. But he
did the rest of us an invaluable service by wrapping up this circle of ideas in
such an entertaining, informative, and easily digestible package. Highly rec-
ommended.

A good introduction to the idea of a Turing machine is given in the Hauge-
land book cited earlier. See also
Rucker, R. Infinity and the Mind. Boston: Birkhauser, 1982.

Hilbert’s formalist program for mathematics is based upqn the idea that
mathematics can be viewed as an activity in which we derive certain strings of
symbols from certain other strings of symbols according to a set of rules. To
avoid infinities, Hilbert required that only finitistic methods be used, where a
method is finitistic if it involves no infinite searches and can be specified in a
finite number of steps. It was Hilbert’s view that one could find a finitist proof
of the consistency of mathematics. As noted in the text, Godel’s Incompleteness
Theorem shattered this illusion once and for all by showing that not only is
any given formal system incomplete, but that there is ho finitely given formal
system that can prove all true statements about the arithmetic of real numbers.
A good, but slightly technical, reference on these matters is
Webb, J. Mechanism, Mentalism, and Metamathematics. Dordrecht, Nether-
lands: Reidel, 1980.

Somewhat less technical introductions to Hilbert’s program, as well as to
Godel’s results, are the Hofstadter and Rucker books discussed above, as well
as

Wang, H. From Mathematics to Philosophy. New York: Humanities Press,
1974,

from which the Godel quote regarding the possibility for thinking machines
was taken. For an account of Chaitin’s work on information-theoretic versions
of Godel’s Theorem as well as much more on the relationships of machines,
formal systems, computability, and biology, see the collection
Chaitin, G. Information, Randomness and Incompleteness. Singapore: World
Scientific, 1987.

Also of interest in this same connection is the more popular treatment in
Rucker, R. Mind Tools. Boston: Houghton-Mifflin, 1987.

"STRONG" VS. "WEAK" AI, BRAINS, AND MINDS
An excellent discussion of the origins and goings-on at the pioneering Dart-
mouth summer gathering is provided by McCorduck in the volume cited under
General References. It makes particularly interesting reading to look at the
interviews with McCarthy, Minsky, Simon, and others present at the meeting,
comparing their feelings at the time about the future course of AI with the
way things have actually worked out.

534

TO DIG DEEPER

The various categories of “strong” and “weak” AI are discussed in consid-
erably more detail in

Gunderson, K. Mentality and Machines, 2nd Edition. Minneapolis: University
of Minnesota Press, 1985.

Gunderson provides not only a useful categorization for sharpening the “Can
machines think?” question, but also offers an extremely thought-provoking cri-
tique of Turing’s Imitation Game. On the fundamental question, Gunderson
concludes that without addressing the mind-body relationship, no progress is
possible on strong AI. But to do this, we need a first-person perspective to be
somehow encoded into an essentially third-person set of descriptions.

As part of his work on the theoretical foundations of computing and ma-
chines, von Neumann discovered that there was no theoretical barrier to the
idea of a self-reproducing machine. Further, he showed that such a machine
would necessarily have to contain an encoded description of itself, i.e., he capa-
ble of self-reference in some definite sense. Thus, it’s doubly odd that he
seemed so pessimistic about the idea of a computer duplicating the cognitive
powers of the human brain. Von Neumann’s final (unfinished) work, in which
he lays out some of his thoughts on the matter, is the text of his Silliman
Lectures:

von Neumann, J. The Computer and the Brain. New Haven, CT: Yale Univer-
sity Press, 1958.

Another excellent volume exploring the brain-mind-machine connection is
Arbib, M. Brains, Machines, and Mathematics, 2nd Edition. New York:
Springer, 1987.

TOP-DOWN SYMBOL CRUNCHING
The treatment given here of the underlying principles of the Simon and Newell
top-down programs really doesn’t do justice to the ideas employed. The Hauge-
land and McCorduck books cited above provide a balanced historical and semi-
technical view. But as always in matters of this sort, it’s preferable to hear
from the protagonists themselves. For this, the best nontechnical introduction
is the classic book

Simon, H. The Sciences of the Artificial, 2nd Edition. Cambridge, MA: MIT
Press, 1981.

For an introductory but illuminating account of SHRDLU, see Hofstadter’s
magnum opus. The quote by Winograd can be found in
Waldrop, M. “Machinations of Thought.” Science ’85, March 1985, p. 44.

Schank gives a popular account of his work on scripts in

Schank, R., and P. Childers. The Cognitive Computer: On Language, Learning,

and Artificial Intelligence. Reading, MA: Addison-Wesley, 1984.

For a detailed blow-by-blow record of the development of a script-following
program in Wilensky’s lab at Berkeley, as well as a firsthand account of the
battles between the Alers and Dreyfus-Searle, see
Rose, F. Into the Heart of the Mind: An American Quest for Artificial Intelli-
gence. New York: Harper and Row, 1984.

TO DIG DEEPER

535

An illuminating account of the problems of getting computers to “understand”
is given in

Winograd, T., and P. Flores. Understanding Computers and Cognition. Read-
ing, MA: Addison- Wesley, 1986.

BOTTOM-UP EMERGENCE

The first salvo fired in Hofstadter’s bottom-Up AI program was his Oddel,
Escher, Bach book cited earlier. Subsequently, he put down his philosophy on
the dream underlying mainline, top-down AI and his objections to it in the
paper

Hofstadter, D. “Waking Up from the Boolean Dream, or, Subcognition as
Computation,” in Metamagical Themas, pp. 631-665. New York: Basic, 1984.

For details about the principles on which Jumbo is based, along with additional
information about its inner workings, see
Hofstadter, D. “The Architecture of Jumbo. ” Proceedings of the 2nd Machine
Learning Workshop, 1983, pp. 161-170.

The problems of identifying letterforms, as well as those involved in trying to
get a program to do analogies, are discussed in more detail in the Metamagical
Themas volume. A popular account of the Hofstadter position vis-a-vis “classi-
cal” AI, together with a consideration of the competing personalities as well as
their programs, is found in

Gleick, J. “Exploring the Labyrinths of the Mind.” The New York Times
Magazine, August 21, 1983, p. 23.

The flavor of the guerrilla warfare being waged between the competing AI
schools is captured in the acerbic commentary by Newell on Hofstadter’s views
reported in

The Study of Information: Interdisciplinary Messages, F. Machlup and U.
Mansfield, eds. New York: Wiley, 1983.

Hofstadter’s rejoinder can be found in the postscriptum to his “Boolean
Dream” article referenced above.

For a detailed view of Marvin Minsky’s thinking on mental kollectivs, see
his book

Minsky, M. The Society of Mind. New York: Simon and Schuster, 1987.

Also of interest is the Minsky and Papert treatment of perceptrons, which is
reported in the following new edition written to take account of the revival of
the perceptron idea in the new connectionism:

Minsky, M., and S. Papert. Perceptrons, Enlarged Edition. Cambridge, MA:
MIT Press, 1988.

Lenat’s evolutionary approach to bottom-up cognition is discussed in the
Waldrop work cited earlier.

A popular introduction to the general philosophy and program of the new
connectionists is

“Seeking the Mind in Pathways of the Machine.” The Economist, June 29,
1985, p. 83.

536

TO DIG DEEPER

A vastly more detailed, technical account of the entire effort is given in

Parallel Distributed Processing, Yol. 1: Foundations, Vol. 2: Psychological and
Biological Models, J. McClelland and D. Rumelhart, eds. Cambridge, MA:
MIT Press, 1986.

PHILOSOPHERS AGAINST: THEY'LL NEVER THINK!

An excellent paper summarizing all the arguments of the “Computers can’t
think” crowd is

Grabiner, J. “Artificial Intelligence: Debates About Its Uses and Abuses.”
Historica Mathematica, 11 (1984), 471-480.

Details of the Dreyfus arguments against the AI community are given in the
volumes

Dreyfus, H. What Computers Can’t Do: The Limits of Artificial Intelligence,
Revised Edition. New York: Harper Colophon, 1979.

Dreyfus, H., and S. Dreyfus. Mind over Machine. New York: Free Press,
1986.

Many more details of the historical development of the Dreyfus case, along
with extensive commentary and interviews with his opponents, are found
in McCorduck’s book cited above. For a more specific, technically based
attack, see

Wilks, Y. “Dreyfus’ Disproofs.” British Journal for the Philosophy of Science,
27 (1976), 177-185.

The original reference for Lucas’s argument from Godel is

Lucas, J. “Minds, Machines, and Godel.” Philosophy, 36 (1961), reprinted in

Minds and Machines, A. Anderson, ed. Englewood Cliffs, NJ: Prentice-Hall,

1964.

Objections to Lucas are put forward in Hofstadter’s Godel, Escher, Bach. More
technical arguments are given in

Benacerraf, P. “God, the Devil, and Godel.” The Monist, 51 (1967), 9-32.

Searle’s Chinese Room-style arguments against strong AI are amplified in
his Reith Lectures given on the BBC. These lectures have been published as
the book

Searle, J. Minds, Brains, and Science. Cambridge, MA: Harvard University
Press, 1984.

The extreme generality of the Turing Test, together with a spectrum of en-
tertaining arguments supporting its claim as an indicator of intelligence, is
explored by one of the leading philosophers supporting AI in
Dennett, D. “Can Machines Think?” in How We Know, M. Shafto, ed. New
York: Harper and Row, 1985.

THE MORALIST AND THE MYSTIC

A detailed summary of Weizenbaum’s arguments involving the humanity-ver-
sus-machine issue is found in

Weizenbaum, J. Computer Power and Human Reason. San Francisco: Free-
man, 1976.

TO DIG DEEPER

537

As far as I can tell, no one has been convinced of the case Weizenbaum tries to
make and his arguments are seldom heard any longer. However, when the book
was first published there were many impassioned discussions about the points
raised. For an account of these, see the McCorduck book already cited.

Rucker’s mystical views are outlined in his book noted above, as well as in
the paper

Rucker, R. “Towards Robot Consciousness.” Speculations in Science and Tech-
nology, 3 (1980), 205-217.

BRINGING IN THE VERDICT

For a somewhat more detailed, technical account of my views expressed here
on matters of self -reference as it pertains to a system’s ability to contain model
of itself, as well as my contentions about the distinction between a model and a
simulation, the interested reader should consult Chapter Seven of
Casti, J. Alternate Realities: Mathematical Models of Nature and Man. New
York: Wiley, 1989.

CHAPTER SIX
GENERAL REFERENCES

Numerous popular and semipopular treatments of the ETI question have been
published in recent years, examining the topic from various points of view.
Here is one of the best:

Shklovskii, J., and C. Sagan. Intelligent Life in the Universe. San Francisco:

Holden-Day, 1966.

This volume really kicked off the SETI era in several ways. First of all, it is a
thorough, scientifically documented, and literate account of all aspects of the
SETI question, circa the mid-sixties. Furthermore, the book represents a
unique kind of collaboration between the Russian Shklovskii and the American
Sagan, which originally began as just a translation of a similar book in Rus-
sian by Shklovskii, but turned into a major collaborative venture on what
amounts to a different book. With the exception of some of the experimental
work, most of the material covered is still relevant today and can be read with
profit. Highly recommended.

A more recent account of theoretical and experimental ETI is given in
semipopular form by

Baugher, J. On Civilized Stars: The Search for Intelligent Life in Outer Space.

Englewood ClifEs, NJ: Prentice-Hall, 1985.

James Trefil is a physicist at George Mason University in Virginia, and a
man well known for his popular books on the wonders of physics and Nature.
His colleague Robert Rood is an astronomer interested in ETI. Over a few
beers at a local pub they put their heads together and started speculating
about the ETI question, trying to address the issues from as unbiased a view-
point as possible within the constraints of human prejudice. Their conclusions
(which are not the same for each author) are presented in the popular account

538

TO DIG DEEPER

Rood, R., and J. Trefil. Are We Alonef The Possibility of Extraterrestrial

Civilizations. New York: Scribner’s, 1981.

An excellent source for many of the pioneering ETI papers, as well as a
representative selection of readings outlining various aspects of the ETI ques-
tion, is the collection

The Quest for Extraterrestrial Life: A Book of Readings, D. Goldsmith, ed. Mill

Valley, CA: University Science Books, 1980.

It has been argued that the resolutely anthropomorphic bias of most SETI
work may blind us to how aliens might communicate and think. For a fasci-
nating account of this aspect of SETI from the point of view of a psychol-
ogist, see

Baird, J. The Inner Limits of Outer Space. Hanover, NH: University Press of

New England, 1987.

No topic has provided more ammunition for the science fiction writers than
contact with ETI in all its possible forms. To my eye, some of the best accounts
focusing on radio contact are

Gunn, J. The Listeners. New York: Scribner’s, 1972.

Lem, S. His Master’s Voice. New York: Harcourt Brace Jovanovich, 1983.

McDevitt, J. The Hercules Text. New York: Berkley, 1986.

Sagan, C. Contact. New York: Simon and Schuster, 1985.

These volumes all have the same basic theme: the receipt, translation, and in-
terpretation of a signal, and the way in which human hopes, fears, and interac-
tions are affected by the knowledge that ETI exists. Each of these volumes has
its own answer to the question “What does communication with ETI mean for
mankind?,” my own favorite being the less than gushing account given by
Lem. „

The literature on direct contact is so enormous that it’s impossible to give
even a representative sampling of the many themes that have been explored.
Instead let me list just a smattering of my personal favorites:

Bova, B. Voyagers. New York: Doubleday, 1981.

Crichton, M. Sphere. New York: Knopf, 1987.

Forward, R. The Dragon’s Egg. New York: Ballantine, 1980.

Lem, S. Solaris. London: Faber and Faber, 1971.

McCollum, M. Life Probe. New York: Ballantine, 1983.

Moffitt, D. The Jupiter Theft. New York: Ballantine, 1977.

The usual Hollywood vision of how we would react to the landing of an alien
vessel is something along the lines depicted in Close Encounters, where everyone
is calm, peaceful, and full of cosmic harmony and goodwill. Some observ-
ers, myself included, feel considerably less sanguine about the possibility.
The results of Orson Welles’s Mercury Theatre radio broadcast of The
War of the Worlds on Halloween 1938 suggest that the most likely outcome
of such direct contact will be nothing less than sheer terror. This aspect of
SETI appears to await an enterprising investigator from the psychological
community.

TO DIG DEEPER

539

THE FERMI PARADOX AND PROIECT OZMA
While putting this section together, I wanted to track down as precisely as
possible the moment that SETI entered the experimental phase with the begin-
ning of Project Ozma. It seemed to me that such a significant beacon on the
SETI landscape would be well chronicled, especially since it happened only
a couple of decades ago. As an indication of why scientists make poor his-
torians, I found the following dates in 1960 given: April 8 (Baugher, 1985),
April 11 (Papagiannis, 1985), autumn (Shklovskii and Sagan, 1966), May-
June-July (McGowan and Ordway, 1966), early 1960 (Sagan, in a 1974 arti-
cle), spring (Hood and Trefil 1981), and, worst of all, no date at all offered by
Frank Drake himself, in an account of the experiment published only one year
after it had been completed. What a mess! The date of April 11 quoted in the
text is taken from personal accounts of the Ozma search given at a twenty-
fifth-anniversary Fest held at the National Radio Astronomy Observatory, the
proceedings of which are reported in

The Search for Extraterrestrial Intelligence, K. Kellerman and O. Seielstad,

eds. Green Bank, W V: NRAO, 1986.

Living in the modem age of risk-averse science, NSF peer review, and un-
imaginative scientific apple polishing, I find it refreshing to read Drake’s ac-
count of how there was no proposal, no committee, no referees, no studies, just
an OK from NRAO Director Otto Struve. In short, science as it should be —
done by scientists and not by congressmen, NSF program managers, university
veeps, or political action groups.

The classic paper advocating the 1420-MHz “waterhole” frequency as the
natural place to look for ETI is

Cocconi, G., and P. Morrison. “Searching for Interstellar Communications.”

Nature, 184 (1959), 844.

Plans for Project Ozma and the publication of the Cocconi and Morrison paper
were progressing along totally independent lines. So when the paper appeared,
NRAO Director Otto Struve was apparently quite agitated, wanting to ensure
that appropriate credit for the idea of a search would go not to Cocconi and
Morrison, but rather to the newly founded NRAO. As a preemptive strike,
Struve totally changed a talk scheduled the following week at MIT to empha-
size the Ozma project, thereby putting the NRAO on record with the idea.
Struve was clearly a man with a deep understanding of the ways of credit in
science, not to mention the bureaucratic one-upsmanship needed to keep a
fledgling organization visible where it counted — with the funding agencies.

THEORETICAL ETI: THE DRAKE EQUATION
The Drake equation was first formulated at a meeting in November 1961 at the
National Radio Astronomy Observatory, only a year after the Ozma search.
Since then many alternate formulations have been offered, although the key
astrophysical, biological, and sociocultural components have remain un-
changed.

One of the major objections to the use of the Drake equation in ETI studies

540

TO DIG DEEPER

is that each of its components can itself be decomposed into an almost endless
list of “sub-Drake” equations. For example, the term f„ involving the likeli-
hood of the appearance of life, lumps into one number a large collection of
separate steps, each of which has its own probability of occurrence. Carrying
out this kind of microanalysis on each of the terms leads to a “super” Drake
equation containing about as many terms as you wish. If each such term has a
likelihood of less than one, multiplying them together creates an estimate for N
that is as low as your prejudices require. The counterargument to the resulting
claim that the Drake equation is useless is to claim that the arbitrarily small
estimates of N arise from the assumed independence of the individual terms. If
some are dependent, then all bets are off and the equation can again be em-
ployed. More details on all these matters are found in the NRAO volume cited
above.

SLICES OF THE ETI PIE

Far more detailed accounts of the various slices of the ETI pie are found in
the Sagan and Shklovskii, Rood and Trefil, and Baugher volumes noted above.

The simulations of possible planetary systems are taken from

Dole, S. “Computer Simulation of the Formation of Planetary Systems.”

Icarus, 13 (1970), 494-508.

Hart’s calculations showing the narrow path that the Earth had to tread
in order to avoid becoming either a frozen wasteland or a Turkish bath are
given in

Hart, M. “Habitable Zones About Main Sequence Stars.” Icarus , 37 (1979)

351-357.

More recent calculations show that perhaps the CHZ is not as small as Hart
imagined. These models, based upon the idea that concentrations of carbon di-
oxide in the atmosphere would be enough to prevent water from freezing, even
on planets far from their parent star, push the CHZ for Earth-like planets
from Hart’s estimate of 0.95-1.05 ATT to 0.95—1.5 ATT, an increase of almost 50
percent. For details, see the account

“Model Atmospheres Show Signs of Life.” New Scientist. January 7, 1988,

p. 41.

Discussions of Miller’s classic experiment are found in almost every book on
ETI, this one not excepted. The current guru of this type of investigation
aimed at showing how life could (must?) have arisen on Earth is Cyril Pon-
namperuma of the University of Maryland. A good account of the present
state of this arcane chemical art is

Ponnamperuma, C. “Primoridial Organic Chemistry,” in Extraterrestrials:

Where Are Theyf, M. Hart and B. Zuckerman, eds. New York: Pergamon

1982.

ANTHROPOMORPHISMS, CHAUVINISMS, AND
ETI NUMEROLOGY

In Table 6.1, the Hart estimates for the value of N are found in
Hart, M. “N Is Very Small,” in Strategies for the Search for Life in the Uni-
verse, M. Papagiannis, ed., pp. 19-25. Dordrecht, Netherlands: Reidel, 1980.

TO DIG DEEPER

541

When interpreting the statistical estimates given by Sturrock, one should be
extremely careful to note that the legitimacy of the conclusions regarding the
confidence levels for N are entirely dependent upon the accuracy of the various
estimates that went into the statistical analyses. Thus, while Sturrock’s statis-
tical wizardry may be beyond reproach, if the raw data regarding N that
formed the basis for his calculations is hopelessly adrift, then so is the credibil-
ity of the final conclusions. Readers interested in a complete account of Stur-
rock’s analysis can find it in

Sturrock, P. “Uncertainty in Estimates of the Number of Extraterrestrial
Civilizations,” in Strategies for the Search for Life in the Universe, M. Papa-
giannis, ed., pp. 59-72. Dordrecht, Netherlands: Reidel, 1980.

The complete transcript of the historic Byurakan meeting, along with a
number of supplementary documents including a discussion of the notion of
subjective probability, is found in the following volume which is must reading
for anyone curious about ETI:

Communication with Extraterrestrial Intelligence, C. Sagan, ed. Cambridge,
MA: MIT Press, 1973.

Informal, personal accounts of the goings-on at this Armenian gathering by
two of the participants are given in

McNeill, W. “Journey from Common Sense.” University, of Chicago Magazine,
64 (May-June 1972), 2-14.

Dyson, F. “Letter from Armenia.” The New Yorker, November 6, 1971,

p. 126.

These accounts illustrate that even such cerebral gatherings are not without
their lighter side, as evidenced at Byurakan by Joseph Shklovskii’s response to
someone’s slightly harebrained idea that there was a strong correlation between
sunspot maximums and the appearance of notable examples of human creativ-
ity. Shklovskii observed that “this theory was obviously concocted during a
period of a deep sunspot minimum!”

Dyson is one of America’s most publicly visible physicists, having been in-
volved not only in pioneering research in quantum theory, hut also serving as a
member of the Orion Project devoted to the design of a low-cost vehicle for
human space travel. In addition, he has been a tireless crusader for a more
sane view of the dangers of uncontrolled nuclear weaponry. His autobiogra-
phy, detailing his feelings on these issues for a general audience, is
Dyson, F. Disturbing the Universe. New York: Harper and Row, 1979.

A far different side of Dyson’s life, one showing that even eminent theoretical
physicists are not immune to the kinds of generational parent-child conflicts
that plague the rest of us, is provided by the candid profile of Dyson and his
son, George, given in

Bower, K. The Starship and the Canoe. New York: Holt, Rinehart and Win-
ston, 1978.

EXPERIMENTAL SETI: HOW SHOULD WE LISTEN?
Up-to-date accounts of the various types of radio searches for ETI are given
in these survey articles:

542

TO DIC DEEPER

Papagiannis, M. “Recent Progress and Future Plans on the Search for Ex-
traterrestrial Intelligence.” Nature, 318 (1985), 135-140.

Tarter, J. “SETI Observations Worldwide,” in The Search for Extraterrestr-
ial Intelligence, K. Kellermann and O. Seielstad, eds., pp. 79-98. Green
Bank, WY: NRAO, 1986.

The NRAO volume also contains a number of other papers outlining the spe-
cific details of a variety of SETI radio searches underway or planned, includ-
ing details of the NASA program.

Dyson’s idea of dismantling one’s home solar system to create a material
sphere surrounding its sun was initially proposed in the one-page note
Dyson, F. “Search for Artificial Stellar Sources of Infrared Radiation.” Sci-
ence, 131 (1960), 1967.

For some strange, inexplicable reason, this idea appears to have captured the
fancy of the Russians, and several Soviet searches have been conducted looking
for such “hot” sources of infrared radiation. Somehow the idea has never
seemed as appealing to American astronomers and, as far as I can tell, it’s
currently on the back burner of U.S. SETI activity.

For an account of Michael Papagiannis ’s arguments for why the asteroid
belt might be a good place to look for ETI, see
Papagiannis, M. “Colonies in the Asteroid Belt, or a Missing Term in the
Drake Equation,” in Extraterrestrials: Where Are They f, M. Hart and B.
Zuckerman, eds., pp. 77-86. New York: Pergamon, 1982.

WHAT ARE WE LISTENING FOR? — THE SYNTAX AND
SEMANTICS OF SETI

A very readable popular account of the entire SETI issue, including a number
of interesting graphics illustrating the communication problem, is given in
McDonough, T. The Search for Extraterrestrial Intelligence: Listening for Life
in the Cosmos. New York: Wiley, 1987.

Pictorial radio messages are not the only type of language that’s been sug-
gested for communicating with ETI. Some years ago, Dutch mathematician
Hans Freudenthal developed a purely logical, nonverbal, semantic-based lan-
guage called LINCOS (for Lingua Cosmica) for such messages to the stars.
While the study of terrestrial languages includes grammar, syntax, and pho-
nemics, LINCOS is designed entirely in terms of semantics. It consists of a
coded system of units that are clearly enumerated as chapters and paragraphs.
This structure facilitates the interpretation of the message, as the semantic
content can be derived from logic external to the linguistic system itself. A
LINCOS transmission begins with the most elementary concepts of mathemat-
ics and logic, since the language must describe itself before it can be used as a
communications medium. After this “self-definition” phase, the language goes
on to logically develop more complicated concepts of the natural, social, and
behavioral sciences. For a detailed description of the language and its use, see
Freudenthal, H. LINCOS: Design of a Language for Cosmic Intercourse. Am-
sterdam: North-Holland, 1960.

TO DIG DEEPER

543

The initial idea for the plaque on Pioneer 10 seems to have come from the
science writer Eric Burgess, who realized that the probe would become the first
human artifact ever to leave the solar system. He got in touch with a colleague,
writer Richard Hoagland, who in turn contacted Carl Sagan, and the rest was
history. As an irrelevant footnote to the whole episode, Sagan’s former wife
Linda Saltzman was responsible for the drawings of the naked male and female
figures that caused all the ruckus about space pornography.

An excellent account of the development of the far more ambitious project to
put a record of Earth on the Voyager probes is given in the McDonough book
cited earlier, as well as the following volume produced by the project team itself:
Sagan, C., P. Drake, A. Druyan, T. Perris, J. Lomberg, and L. Saltzman
Sagan. Murmurs of Earth. New York: Ballantine, 1978.

Here is the solution to Prank Drake’s hypothetical message from ETI:

The picture shows a figure of a humanoid whose home star is given along the
left border surrounded by the nine planets of its solar system. The figure’s

544

TO DIG DEEPER

upper-right corner shows diagrams of carbon and oxygen, suggesting that
the BTI’s body chemistry is similar to our own. Immediately to the right of
the first five planets are shown the first five positive integers, written in bi-
nary fashion with a parity bit, i.e., 1 = 10, 2 = 100, 3 = 111, 4 = 1000, 5
= 1011. Note that the parity bit causes each number to have an odd number
of 1 s. To the right of the alien, and connected by a diagonal line, is a car-
toonist’s balloon inside of which are three numbers (you can tell they’re
numbers because they each have an odd number of l’s). There is the number 5
next to planet 2 ; 868 by planet 3 ; and about 4 billion next to planet 4.
Presumably these numbers reflect the populations of aliens on these planets,
indicating an exploratory expedition on the second planet and a colony on
the third, while planet 4 is the home planet. To the right of the alien is its
height of 31 “units,” which it is logical to assume are the natural units of
the message itself, the wavelength of the transmitting signal. The line of
four blocks underneath the alien itself might be interpreted as the alien’s
code for itself, since it can’t be a number (because it has an even number of
l’s).

Further amplification and elaboration of Ball’s list of possibilities for
ETI is given in his original article:

Ball, J., “Extraterrestrial Intelligence: Where Is Everybody?” in The
Search for Extraterrestrial Life: Recent Developments , M. Papagiannis, ed.,
pp. 483-486. Dordrecht, Netherlands: Reidel, 1985.

N > 1: ETI EXISTS!

The argument given here about the outrageous costs of mounting a manned
exploration of even a nearby star system is given in detail in
Drake, F. “N Is Neither Very Small Nor Very Large,” in Strategies for the
Search for Life in the Universe, M. Papagiannis, ed., pp. 27-34. Dordrecht,
Netherlands: Reidel, 1980.

Other arguments conclude just the opposite, saying that star travel is well
within our projected pocketbook. For an account of these claims, see
Interstellar Migration and the Human Experience, B. Finney and E. Jones,
eds. Berkeley, CA: University of California Press, 1985.

Sagan, C. “Direct Contact Among Galactic Civilizations by Relativistic In-
terstellar Spaceflight,” Planetary and Space Science, 11 (1963), 485.

Singer, C. “Settlements in Space, and Interstellar Travel,” in Extraterrestr-
ials: Where Are Theyf, M. Hart and B. Zuckerman, eds., pp. 46-61. New
York: Pergamon, 1982.

THE SHAPE OF ETIS TO COME

The possibilities for alien anatomy are virtually endless, with the science fic-
tion literature having at one time or another explored most of them. For those
like myself having a congenital weakness for such kinds of speculation, the
following volume of artistic interpretations of ETI is absolutely must reading:
Barlowe, W., and I. Summers. Barlowe’s Guide to Extraterrestrials: Great
Aliens from Science Fiction Literature. Leicester, UK: Windward, 1979.

TO DIG DEEPER

545

Speculations not only on the nature of ETI’s anatomy, but also on social
structures and life-styles, can be found in
Cultures Beyond Earth, M. Maruyama and A. Harkins, eds. New York: Vin-
tage, 1975.

Forward, R. “When You Live Upon a Star . . . New Scientist, December
24, 1987, pp. 36-38.

Jonas, D. and D. Other Senses, Other Worlds. New York: Stein and Day,
1976.

ETI? THERE'S NO SUCH THING: N = 1
For light comic relief, the article below by Adler is tough to beat. For some
low-level laughs, have a look at

Adler, A. “Behold the Stars.” Atlantic Monthly, 234 (1974), 109.

Michael Hart’s pathhreaking ETI paper, trying to show that the emperor
has no clothes, is

Hart, M. “An Explanation for the Absence of Extraterrestrials on Earth.”
Quarterly Journal of the Royal Astronomical Society, 16 (1975), 128-135.

The countervailing claim that “absence of evidence is not evidence of absence”
can be found in

Cox, L. “An Explanation for the Absence of Extraterrestrials on Earth.”
Quarterly Journal of the Royal Astronomical Society, 17 (1976), 201.

Tipler’s classic contribution to the SETI debate was first published in
Tipler, F. “Extraterrestrial Intelligent Beings Do Not Exist.” Quarterly
Journal of the Royal Astronomical Society, 21 (1980), 267-281.

For a personal account of the behind-the-scenes machinations surrounding
publication of the above paper, as well as additional commentary, the reader
should consult Tipler’s contribution to the volume
Rothman, T., et al. Frontiers of Modem Physics. New York: Dover, 1985.

Not to be upstaged by Tipler’s arguments, Carl Sagan, the paper’s original
referee, had lots of time to muster his ammunition against Tipler’s claim,
which he termed the solipsist approach to ETI. See
Sagan, C., and W. Newman. “The Solipsist Approach to Extraterrestrial
Intelligence.” Quarterly Journal of the Royal Astronomical Society, 24 (1983),
113-121.

The anthropic argument given by Carter for the nonexistence of ETI is most
easily accessed in

Barrow, J., and F. Tipler. The Anthropic Cosmological Principle. Oxford: Ox-
ford University Press, 1986.

Accounts of Simpson’s biological objections to ETI can be found reprinted
in the Goldsmith volume cited above under General References, while Mayr’s
arguments are given in the volume

Extraterrestrials: Science and Alien Intelligence, E. Regis, ed. Cambridge:
Cambridge University Press, 1985.

546

TO DIG DEEPER

This volume, incidentally, contains a wealth of additional material on all sides
of the SETI question and is highly recommended as a general reference.

Nicholas Eescher’s arguments for the likelihood of ETI’s science being
“weird” by our standards can be found in “Extraterrestrial Science,” pp. 83-
116 in the Regis book cited above. In this same connection, see Regis’s article
“SETI Debunked,” pp. 231-244 in the same volume.

For the arguments by Eigen, Schuster, and Dawkins regarding the ability of
complex systems to form “randomly” using the ratcheting principle, see their
popular works:

Eigen, M., and P. Schuster. The Hypercycle: A Principle of Self-Organization.

Berlin: Springer, 1979.

Dawkins, R. The Blind Watchmaker. London: Longman, 1986.

SUMMARY ARGUMENTS

In Table 6.4, I have noted Freeman Dyson’s argument for comets as a likely
home for ETI. While this doesn’t exactly constitute a claim that ETI exists,
it’s an intriguing idea for getting around in the universe: Just hitch a ride on
a comet and your energy problems are over since you can let Nature can pay
the bill. Dyson’s principal claim is that since there are a lot of comets around,
each of which contains an abundance of free raw material, this would be a
likely way for a cost-conscious ETI to go if it wanted to look over the galaxy —
provided it wasn’t in a big hurry!

CHAPTER SEVEN
GENERAL REFERENCES

Bookshops are literally overflowing with volumes at all degrees of technical
sophistication offering to “explain” the paradoxes of the quantum world to the
uninitiated. Many of them do a pretty good job; some are misleading; others
are just artless junk. Among the works in the first category, one stands out in
my mind as being the hands-down winner when it comes to a thoroughly read-
able, highly enlightening, vastly entertaining, well-illustrated nontechnical
treatment of quantum mischief. That volume, upon which I have shamelessly
modeled some of the earlier sections of this chapter, is

Herbert, N. Quantum Reality: Beyond the New Physics. New York: Double-
day, 1985.

There seems to be something about quantum theory that brings out the poet in
writers who attempt to convey the ideas to the general reader. In addition to
the Herbert book above, three other accounts highly recommended for the not
especially technically inclined are

Pagels, H. The Cosmic Code. New York: Simon and Schuster, 1982.

Rae, A. Quantum Physics: Illusion or Reality f Cambridge: Cambridge Univer-
sity Press, 1986.

Squires, E. The Mystery of the Quantum World. Bristol, UK: Hilger, 1986.

TO DIG DEEPER

547

In the last decade or so, it has become a bit of a fad to try to link quantum
reality as discussed here with all sorts of mystical ideas having their roots in
various Eastern religions. While I hold no particular brief for these efforts, I
feel that, like anyone attempting to capture the uncapturable, some of the au-
thors do a better job of hitting the target than others. A volume worthy of
honorable mention in this connection is

Zukav, G. The Dancing Wu Li Masters. New York: Morrow, 1979.

Two volumes of the “quantum theory *— • mystical world” type that in my
opinion a discriminating reader can safely miss are

Toben, B., and F. Wolf. Space-Time and Beyond. New York: Dutton, 1975.
Wolf, F. Star Wave. New York: Macmillan, 1984.

Both of these volumes (especially the second) are of the sort that contribute to
the enthusiasm with which most physicists regard quantum reality research as
being a highly suspect activity, if not downright unscientific or even unprofes-
sional. Strangely enough, author Fred Wolf is a trained physicist whose ear-
lier book Taking the Quantum Leap won the American Book Award for science
exposition. That effort was, in my view, a successful attempt to explain the
concepts and principles of the quantum world to a general audience. However,
with Star Wave, a fairly evident attempt to reach an even wider audience, the
author flies off the track with a host of outrageous speculations about quantum
theory and its relevance to new laws of psychology, love, hate, sanity, mind
control, death, reincarnation, and a whole lot more. While this sort of thing
probably does sell books, it doesn’t do much to further the understanding of
the limitations of quantum theory for curing the worlds ills. While I’d never
endorse any kind of “ban the book” initiative, I would feel more comfortable if
books like this were not around.

In a more positive vein, the history of both the ideas and the people of quan-
tum mechanics is vividly portrayed in the following works:

Cline, B. Men Who Made a New Physics. Chicago: University of Chicago
Press, 1987.

Jammer, M. The Philosophy of Quantum Mechanics. New York: Wiley, 1974.

The Jammer volume is fairly technical in parts, but gives an insider’s account
of the back room discussions, as well as the personality factors, that underlie
how the Copenhagen Interpretation came to quantum ascendancy. The Cline
book is a purely nontechnical version of the same people and events, written in
a clear, informative fashion by a science writer. Both books are to be highly
praised for the light they shed on the human factor in the creation of a scien-
tific revolution.

BUILDING THE STAGE

A more detailed account of Wheeler’s “contextual” twenty-questions game, as
well as a nice, compact introduction to the basic ideas of quantum theory, is
found in the first part of

The Ghost in the Atom, P. Davies and J. Brown, eds. Cambridge : Cam-
bridge University Press, 1986.

548

TO DIG DEEPER

This is an extremely interesting little book, the bulk of which is a collection of
transcripts of interviews originally broadcast on BBC radio with many of the
main actors in the modern quantum-reality game, including Wheeler, Bell, and
Bohm.

GHOSTS IN THE ATOM

For a lavishly illustrated and somewhat more detailed discussion of the double-
slit experiment, the following popular treatment is hard to beat:

Hey, T., and P. Walters. The Quantum Universe. Cambridge: Cambridge
University Press, 1987.

For the overall idea using waveform families, prisms, and the like to explain
the Schrodinger solution to the Description Problem, I am indebted to the out-
standing treatment provided in Herbert’s book cited above. The reader is
strongly encouraged to consult Herbert for a more leisurely account. Inciden-
tally, to be perfectly accurate, the quantities termed c, in the text are related to
the actual values of the quantum wave function W(x, <), which is complex-
valued. This is necessary for W(x, <) to display the needed wavelike behavior.
Thus the elements c, are not real numbers but complex quantities, implying
that when we compute the probabilities of various experimental outcomes, we
should really use c,c , = |c, |2, where | ' | is the complex modulus, and not the
simpler c ? of the text. A good source for a proper discussion of these matters
is the well-known textbook

Feynman, R., R. Leighton, and M. Sands. The Feynman Lectures on Physics,
Yol. III. Reading, MA: Addison-Wesley, 1965.

MEASUREMENT TO MEANING

The quote in the text illustrating the kind of misinformation in circulation
regarding the Heisenberg Uncertainty Principle was taken from An Incomplete
Education by J. Jones and W. Wilson (New York: Ballantine, 1987), p. 489.
Here’s another from a different source:

It seems to me that we can apply the Heisenberg uncertainty principle to
the problem of the meaning of words. Writers, poets, etc. use words in a
very large, general sense, but for them they have a very special meaning.
The single word has a very special function in their description. In con-
trast, in the sciences words are very sharply defined and have a very short-
range validity. But this fact made it possible that this word is understood
universally. Restricting the domain of validity of the word produces, on
the other hand, a gain in universality.

This statement, made by a physicist at an interdisciplinary meeting aimed at
bringing scientists, writers, musicians, and others together, might (by a chari-
table interpretation) be thought of as an appeal to the Heisenberg Uncertainty
Principle as a metaphor. But surely the author cannot be claiming that a word
used in a specialized sense is in any meaningful way “conjugate” to that same
word used in an everyday manner. To my mind, it’s an open question whether
or not use of Heisenberg’s principle in this kind of metaphorical sense helps or
hinders the process of bringing science back into contact with the mainstream
of intellectual life.

TO DIG DEEPER

549

THE ROMANTIC REALITIES

To cover the evolution of thought from Copenhagen to Austin and way sta-
tions in between, the following collection of reprints and commentary is must
reading:

Quantum Theory and Measurement, J. A. Wheeler and W. Zurek, eds. Prince-
ton, NJ: Princeton University Press, 1983.

Of special interest in this volume is a series of papers and lectures detailing
the ongoing battle between Einstein and Bohr over the adequacy of the Copen-
hagen Interpretation.

While von Neumann’s “Cut Theorem” appeared to have pulled the rug out
from under the naive realists, it should be kept in mind that von Neumann was
a mathematician, not a physicist. As a result, the assumptions he made in his
quantum bible were the kind that led to a mathematically elegant theory, but
not ones that in retrospect necessarily appear to be physically appropriate. In
fact, John Bell in a recent interview went so far as to call von Neumann’s
proof “silly.” But to illustrate the Great Man Theory of science, von Neu-
mann’s immense prestige as a mathematician convinced the physicists that it
must be so if von Neumann said it, thus setting quantum reality research back
at least thirty years. For those eager to see what the nature of these dubious
assumptions are, the English version of the bible should be consulted:
von Neumann, J. Mathematical Foundations of Quantum Mechanics, R. Beyer,
trans. Princeton, NJ: Princeton University Press, 1955.

For a wide range of ideas about the ways in which science might shed some
light on the problem of consciousness, see the following volume, which reports
the proceedings of a meeting on the topic involving such luminaries as Bohm,
Fritjof Capra, and Nobel laureate Brian Josephson, meeting with a group of
French and Spanish thinkers on the matter:

Science and Consciousness: Two Views of the Universe, M. Cazenave, ed. Oxford:
Pergamon, 1984.

Schrodinger originally put forth his cat paradox in 1935 in the German
journal Naturwissenschaften. An English translation appeared in 1955, coinci-
dentally with the English version of von Neumann’s book. An easily accessible
source for the Schrodinger paper is the Wheeler and Zurek compendium al-
ready cited. For a full treatment of Wigner’s views on the quantum reality
issue, as well as his always insightful reflections on mathematics, physics, and
their mutual dependence, see the collection of papers and essays
Wigner, E. Symmetries and Reflections. Bloomington, IN: Indiana University
Press, 1967.

Wheeler has been a tireless campaigner for the measurement option view of
quantum reality, having written numerous articles and books all hammering
home the idea that observers have a choice in creating the kind of reality they
see. Two good summaries of his ideas are given in
Wheeler, J. A. “Beyond the Black Hole,” in Some Strangeness in the Propor-
tion, H. Woolf, ed., pp. 341-375. Reading, MA: Addison-Wesley, 1980.

550

TO DIG DEEPER

Wheeler, J . A. “How Come the Quantum?,” Annals of the New York Acad-
emy of Sciences, Vol. 480, pp. 304-316. New York: New York Academy of

Sciences, 1986.

A good account of an Earth-based Delayed-Choice Experiment using mir-
rors and light beams is found in the Squires book cited under General Refer-
ences.

Like Schrodinger, Heisenberg was much concerned with the philosophical
implications of quantum theory, including his own Duplex Interpretation. In
his later years, Heisenberg published a number of volumes of essays outlining
his reflections on these and other matters. One of the best is

Heisenberg, W. Physics and Beyond. New York: Harper and Row, 1971.

Borges is far from the only writer who has mined the seemingly inexhausti-
ble vein of “alternative realities” for material to entertain his readers. In my
view, two of the best efforts in this direction by the sci-fi fraternity are

Hogan, J . The Proteus Operation. New York: Bantam, 1985.

Moore, W. Bring the Jubilee. New York: Fantasy House, 1952.

Both of these books deal with “what ifs,” involving branches of the universe in
which the Confederacy (for Moore) or the Nazis (for Hogan) won their respec-
tive wars. I won’t spoil the reader’s enjoyment by giving away the plots, other
than to say that they both involve the usual twists of time travel to set things
“straight” somehow. All in all, good fun.

On the side of sober science, the best source for Everett’s work is the volume

The Many-Worlds Interpretation of Quantum Mechanics, B. de Witt and N.

Graham, eds. Princeton, NJ : Princeton University Press, 1973.

In addition to reprints of Everett’s key papers, this volume also includes an
assessment of the idea by Wheeler as well as a more introductory account by
de Witt. For an account of Deutsch’s version of the MWI, together with a
discussion of an experiment that at least in principle would allow us to make
contact with such worlds, see the BBC interview volume edited by Davies and
Brown cited above.

THE DOC WO R K REALITIES

Einstein s objections to the romantic realists have been chronicled in virtually
every one of the thousands of accounts of his life and times. Generally these
accounts introduce Einstein’s naive realist views by quoting his famous re-
mark, “Clod does not play dice with the universe,” or words to that effect. In
my opinion the best statement of Einstein’s thoughts is, of course, from Ein-
stein himself. It is reported in his autobiography, which forms the first part of
the volume

Albert Einstein: Philosopher-Scientist, Yol. 1, P. A. Schilpp, ed. Lasalle, IL:

Open Court, 1949.

The quantum-logical explanation for the Three-Polarizer Paradox is well ex-
plained in the Herbert volume noted under General References. For a fine dis-

TO DIG DEEPER

551

cussion of the entire quantum-logic idea using only a small amount of under-
graduate mathematics, the reader is urged to consult

Gibbins, P . Particles and Paradoxes: The L/imits of Quantum Logic. Cambridge:
Cambridge University Press, 1987.

A completely nontechnical overview of the Quantum Potential Interpreta-
tion is provided by the editors in the introduction to the following volume of
essays in honor of David Bohm upon his retirement. The introduction traces
the development of Bohm’s thinking on the matter from his first days in
Princeton to his current ideas on the holographic universe. This account is
followed by a slightly more technical discussion by Bohm himself, as well as a
number of papers of varying degrees of difficulty by other heavies such as Bell,
Feynman, and Finkelstein. All in all, a volume to be highly prized, praised,
and perused:

Quantum Implications: Essays in Honor of David Bohm, B. Hiley and F. David
Peat, eds. London: Routledge and Kegan Paul, 1987.

Many of Bohm’s philosophical ideas underpinning the Quantum Potential In-
terpretation are covered in the book

Bohm, D. Causality and Chance in Modem Physics. Philadelphia: University
of Pennsylvania Press, 1957.

For those interested in Bohm’s current thinking about the “holographic uni-
verse,” the following collections of interviews should prove illuminating:
Dialogues with Scientists and Sages, R. Weber, ed. London: Routledge and
Kegan Paul, 1986.

The Holographic Paradigm, K. Wilber, ed. Boulder, CO: Shambhala, 1982.

A historical account of the origin of the quantum potential idea is given by de
Broglie in

Broglie, L. de. “Interpretation of Quantum Mechanics by the Double Solu-
tion Theory.” Annales de la fondation Louis de Broglie, 12 (1987), 399-421.

The original sources for the Absorber Theory are two papers by Wheeler
and Feynman in Reviews of Modem Physics in 1945 and 1949. The modern in-
carnation of the theory according to Cramer is briefly described in the popular
article

Cramer, J. “The Alternate View: The Quantum Handshake.” Analog Science
Fact/Fiction, November 1986.

More technical treatments are given in
Cramer, J. “An Overview of the Transactional Interpretation of Quantum
Mechanics.” International Journal of Theoretical Physics, 27 (1988), 227-236.
Cramer, J. “The Transactional Interpretation of Quantum Mechanics.” Re-
views Modem Physics, 58 (1986), 647-687.

THE BELL TOLLS FOR LOCALITY

I am indebted to Euan Squires’s treatment in his book cited under General
References for the idea of the Alexander and Anastasia telepathy experiment
to illustrate the concepts behind Bell’s Theorem. For another kind of story
illustrating the same principles using flashing colored lights, see

552

TO DIG DEEPER

Mermin, D. “Is the Moon Really There When Nobody Looks! Reality and
the Quantum Theory.” Physics Today, April 1985, pp. 38-47.

The famous EPR Paradox is described in virtually every introductory treat-
ment of the quantum reality question, including all the volumes listed under
General References. The original paper can be found in the Wheeler and Zurek
collection noted above.

An especially good elementary account of the derivation of the Bell inequal-
ity using the colorful image of a nail gun instead of our electron-pair genera-
tor is given by Pagels in his book noted under General References. For a
treatment by the master himself, see the original papers, which are reprinted
both in the Wheeler and Zurek volume and in
Bell, J. S. Speakable and Unspeakable in Quantum Mechanics. Cambridge:
Cambridge University Press, 1988.

Bell s recollections about the origin of his theorem, as well as his thoughts on
Eastern religions, von Neumann’s proof, and current trends in quantum real-
ity, are all reported in

“Interview with John Bell,” Omni, May 1988, 85ff.

A slightly technical but still eminently readable discussion of the Aspect ex-
periments, Bell s Theorem, and the inviability of any local hidden-variable
kind of reality is

Rohrlich, F. “Facing Quantum Mechanical Reality.” Science, 221 (Septem-
ber 23, 1983), 1251-1255.

A popular-science account of the Bell result is given in
d’Espagnat, B. “The Quantum Theory and Reality.” Scientific American, 241
(November 1979), 128-140.

Finally, a fairly technical reference addressing hidden variables and all the
problems of quantum realism is

Redhead, M. Incompleteness, Nonlocality, and Realism. Oxford: Oxford Uni-
versity Press, 1987.

One special point worthy of note about the foregoing volume is its treatment
of the so-called Kochen-Specher Paradox. The essence of this additional quan-
tum paradox is that on the one hand, common sense (again!) would lead us
to expect that the algebraic structure of the operators representing
attributes should be mirrored in the algebraic structure of the set of attri-
bute values themselves. But if this kind of “mirroring” holds, then, Kochen
and Specher show, it is impossible to assign values to all attributes in all quan-
tum states.

IN THE BEGINNING, THE VERY BEGINNING
An entertaining and informative account of the Wilson-Penzias discovery of
the “whispers of the the cosmos” is given in the following treatment of the men
and the science at Bell Labs:

Bernstein, J. Three Degrees Above Zero. New York: Scribners, 1984.

TO DIG DEEPER

553

For an introductory treatment of those first few moments of the universe
after the mysterious origin, there’s no better source than
Weinberg, S. The First Three Minutes. New York: Basic, 1977.

Two very readable discussions of the Eddington-Dirac “numerology” discov-
eries are

Carr, B., and T. Rothman. “Coincidences in Nature and the Hunt for the
Anthropic Principle,” in Frontiers of Modem Physics, T. Rothman et al., eds.,
pp. 108-130. New York: Dover, 1985.

Rothman, T. “A ‘What You See Is What You Beget’ Theory.” Discover,
May 1981, pp. 90-99.

The definitive treatment of all aspects of the anthropic principles is
Barrow, J., and F. Tipler. The Anthropic Cosmological Principle. Oxford: Ox-
ford University Press, 1986.

Many of the topics that have occupied us in the preceding chapters, including
the origin of life, quantum reality, the existence of ETI, and much, much
more, are examined in detail from the anthropic perspective in this seven hun-
dred-page treatise. While the discussion may be a bit too technical for the gen-
eral reader in places, there’s so much material in this encyclopedic volume that
everyone will find something to justify the cost of the book. It’s truly a “don’t
miss it” kind of volume. Less breathtaking, but still excellent, accounts of an-
thropic ideas for the general reader are given in
Boslough, J. Stephen Hawking’s Universe. New York: Morrow, 1985.

Gale, G. “The Anthropic Principle.” Scientific American, December 1981, pp.
114-122.

Greenstein, G. The Symbiotic Universe. New York: Morrow, 1988.

Leslie, J. “Anthropic Principle, World Ensemble, Design.” American Philo-
sophical Quarterly, 19 (1982), 141-151.

Rees, M. “The Anthropic Universe,” New Scientist, August 6, 1987, pp.
44-47.

Critics of anthropic reasoning have put forward a spectrum of reasons why
such ideas have no place in real physics. Steven Weinberg, for example, says
that “I certainly wouldn’t give up attempts to make the anthropic principle
unnecessary by finding a theoretical basis for the value of all the constants.
It’s worth trying, and we have to assume that we shall succeed, otherwise we
surely shall fail.” A somewhat less delicate critique is
Pagels, H. “A Cozy Cosmology.” The Sciences, 25, No. 2 (1985), pp. 34-38.

In this article, Pagels notes that Dicke himself now thinks that the anthropic
principles are worthless unless there was an element of arbitrariness in the
origin of the universe. The argument is simple: If the values of the fundamen-
tal constants were fixed by the laws prevailing at the beginning, then the ques-
tion of the origin of life was settled at the outset and the anthropic principles
are unnecessary. But if there is some randomness in the way the constants are
set, then Dicke thinks the anthropic-style reasoning may have some utility
after all.

554

TO DIG DEEPER

A thorough discussion of the quantum cosmology question is given in the
Barrow and Tipler book noted above. For introductory accounts of how the
universe could have arisen out of nothing more than a quantum fluctuation in
the vacuum, see

Tryon, E. “Is the Universe a Vacuum Fluctuation?” Nature, 246 (1973), 396.
Vilenkin, A. “Creation of the Universe from Nothing.” Physics Letters,
B117 (1982), 25.

Padmanabhan, T. “Quantum Cosmology— Science of Genesis?” New Scien-
tist, September 24, 1987, pp. 60-63.

Speculations from a scientific standpoint on the final state of the universe
appear to be of rather recent vintage, one of the original papers being
Dyson, F. “Time Without End: Physics and Biology in an Open Universe.”
Reviews of Modem Physics, 51 (1979), 447.

A rather thorough discussion of this fascinating topic, emphasizing, of course,
the anthropic perspective, is given in Barrow and Tipler.

INDEX

accommodation, 240
adaptation, 149, 154, 175, 177
adaptive traits, 149
addition, 271-272
Adler, Alfred, 397-398
advanced waves, 465
AI, 286-288, 495-496
and antibehaviorism, 322-324
bottom-up school of, 299-309
connectionist school of, 309-314
morality of, 325-328
and phenomenology, 315-320
strong, 286-288

top-down school of, 290-299
weak, 286

Aleksander, Igor, 310
alien life forms, 394-397
alphabet, 268, 275
Altmann, Sydney, 90
altruism, 171-172, 195
amino acids, 75, 77, 84-85, 98, 356
left-handed, 85
right-handed, 85
analogy, 176

animal aggression, 156-158
anthropic principles, 479-484

556

INDEX

anthropic principles ( cont . )
arguments against, 482-184
final, 482, 487
participatory, 482, 487
strong, 481

weak, 404, 479-480, 485
anticodon, 78
Arecibo message, 380-382
Aristotle, 16

and logical deduction, 16, 18
and nature of reality, 18, 39
theory of causation, 130
will of, 16

Armer, Paul, 316, 319
Arrhenius, Svante, 116
artificial intelligence, see AI
artificial plants, 515
ASCII code, 277
Aspect, Alain, 474
assimilation, 240
attributes, 417
contextual, 442
dynamic, 417, 440
static, 417, 440

Austin Interpretation, 446-450
Automated Mathematician (program),
306-307

Axelrod, Robert, 200-203

Bacon, Francis, 19
and principle of induction, 19
Ball, John, 386

response to Fermi Paradox, 386
Barrow, John, 487
Bauer, Henry, 67
critique of Velikovsky, 501
behavioral traits, 153
genetically altruistic, 154
genetically selfish, 153
phenotypically altruistic, 153
phenotypically selfish, 153
and sexual selection, 163-165
behaviorism, 233-234, 237
radical, 234, 236
belief systems, 62, 66-67, 125,
506-507

Bell, Jocelyn, 2-5
and Nobel Prize, 5-6

Bell, John, 471, 476
Bell’s Theorem, 471-474
Benacerraf, Paul, 321
Big Bang Theory, 476-477, 483
Bigelow, Julian, 39
Black, David, 348
Blinker, 133
Block, 133, 136
Bloom, Allan, 498
Bloomfield, Leonard, 214
Boas, Franz, 214
Bohm, David, 461-462, 464
Bohr, Niels, 441-443, 456-457
Boltzmann, Ludwig, 48-50
suicide, 49-50
and theory of heat, 49
Boltzmann machine, 310-313
Borges, Jorge Luis, 453-454
Boston Group, see Science for the
People Sociobiology Study
Group

Boyle’s Law, 23-24
Brace well, Ronald, 401
Broglie, Louis de, 462
Byurakan SETI meeting, 367,
387-388

final resolution, 388

Calvin, Melvin, 73
carbon chauvinism, 363
Carter, Brandon, 403, 483-484
arguments against ETI, 403-404
categorical imperative, 188
Cech, Thomas, 90
cell, 76
cytoplasm, 76
eukaryotic, 76, 101-102
diagram of, 77
nucleus, 76
prokaryotic, 76
fossil evidence for, 102
reproduction process, 80-82
cellular parasitism, 101
Central Dogma of Molecular
Biology, 82, 108, 147
Central Dogma of Social and

Behavioral Biology, 147, 188,
518

INDEX

557

Central Problem of Modern
Linguistics, 216
Central Tenet of Human
Sociobiology, 152
Chaitin, Gregory, 279-280
Chaitin’s Theorem, 280
chess, 269

Chinese Room test, 265-267
chirality, 85
chloroplasts, 101
Chomskian theory of language,
219-229

Chomsky, Noam, 218-219, 222
debate with Piaget, 250-253
and linguistics, 230
and psychology, 230
and sociobiology, 252-253
views of Skinner’s radical
behaviorism, 236-237
chromosome, 76
Clauser, John, 474
Clever Hans, 210-211
coacervates, 96-98, 355
Cocconi, Giuseppe, 342
codon, 77, 102

coevolutionary circuit, 178-182
diagram for, 180-181
Colby, Kenneth, 325, 327
Complementarity Principle, 442
complexity, 279-280
computer, 275-278
as formal system, 278
logical unit, 276
memory unit, 276
output unit, 276
program, 276
and souls, 329-330
universal, see Turing machine
viruses, 138-139
Comte, August, 31
and evolution of knowledge, 31
conceptual dependency graph, 291
consciousness, 329, 444 445
continuously habitable zone, 351
controlled experiment, 20
Conway, J. H., 132
cooperative behavior, 199
Copenhagen Interpretation, 441-443
Copemican Principle, 493

Cox, Laurence, 401
Cramer, John, 466
creationism, 122-123
and Arkansas Act 590, 123-124
Creation Research Society, 122, 124
Crick, Francis, 82, 115, 117
Cryer, 395-397
crystal growth, 111-113
crystallizer experiment, 115
culture, 359
emergence of, 359-360
culturgen, 178, 195
Cut Theorem, 445, 549
Cygnans, 394-395

Dartmouth conference on AI,
285-286

Darwin’s Formula, 148
Dawkins, Richard, 151, 175, 195,
204,409

deep structure, 229
Delayed-Choice Experiment, 447-449
diagram of, 448
Dennett, Daniel, 266, 324
Deutsch, David, 455
Dicke, Robert, 479-480
Dictionary Correspondence Theorem,
434

Dirac, Paul, 478-479
dissipation, 486
distributive law of logic, 459
Dixon, Robert, 375
DNA, 76-77, 132, 406-407
double-helix structure, 78
Dole, Stephen, 348
Domestic Bliss vs. He-man game,
164-166, 183

double-slit experiment, 420-423
with bullets, 420
with electrons, 422-423
with water waves, 420-421
Drake, Frank, 341-342, 375, 382, 387
Drake equation, 343-345
estimates for N, 365
statistical analysis of, 365-367
Dreyfus, Hubert, 315-317, 320, 333
and RAND Corporation, 316-317
Dreyfus, Stuart, 316-317

558

INDEX

Dual-Origin Theory (Double-Origin
Hypothesis), 99-100, 102
Duplex Interpretation, 450-453
Dyson, Freeman, 47, 367-368, 372
on philosophy of science, 47
Dyson sphere, 372-373

Eater, 133, 136
Eddington, Sir Arthur, 478
Eigen, Manfred, 409
Eigen experiment, 89, 104
Einstein, Albert, 415, 419, 456-
457

electron spin, 431-432
ELIZA (program), 325-326
empirical laws, 22-23
entropy, 49, 304
environment, 148, 154
epigenetic rules, 179-181
EPR Paradox, 470
Bohm version with electrons,
470-471

equilibration, 240
error catastrophe, 94
ETI, 496-497

direct contact via space travel,
389-391

direct visitation by, 391-397
factorization arguments against,
398, 405-409

observation arguments against,
399-404

Eurisko (program), 307
Everett, Hugh, III, 454
evolution, 92
biological, 92, 353
chemical, 92, 353
convergent, 392-393
evolutionary game theory, 158-162
evolutionary stable strategy (ESS),
160, 200-201

Extended General Theory, 244-245
extraterrestrial intelligence, see ETI

falsification (refutability), 33
Feigenbaum, Edward, 290, 319

Fermi Paradox, 340
Feyerabend, Paul, 37-38
and scientific method, 37-38
student experiences of, 504
Feynman, Richard P., 430
finite-state Markov process, 224
Finkelstein, David, 461
fitness, 149-150
genetic, 150, 154
inclusive, 167, 193
in order Hymenoptera, 167-170,
172

maximization, 175
phenotypic, 150, 154
Flanagan, Owen, 185
formal cause, 130-131
formalist program for mathematics,
279, 533

Formal Mode, 130-131
Fox, Sidney, 97-98
frames, 297-299
Fundamental Question of the

Philosophy of Science, 26

Gardner, Martin, 482
Gell-Mann, Murray, 48
gene, 76
regulatory, 76
structural, 76
gene inflation, 175
Gene-Protein Linkup Problem, 84,
89

generative semantics, 245
genetic code, 77, 84
second, 512
table for, 79
translation, 77-79
diagram of, 81

genetic deterioration, 360-361
genetic determinism, 151
genome, 93

genotype, 147-148, 150, 154
Gish, Duane, 124
Glider Gun, 133, 136
diagram of, 135
God, 487

relationship with man, 507

INDEX

559

Godel, Kurt, 279, 281
on thinking machines, 284
Godel sentence, 281
Godel’s Theorems, 279-282, 285
as arguments against AI, 320-322
Gold, Thomas, 387
and pulsars, 500

Gould, Stephen Jay, 182, 187, 192,
206

grammar, 213, 253
decidable, 249
finite-state, 223-225
of formal system, 269, 275
generative, 215, 223
Montague, 246-248
phrase-structure, 223, 225-228
transformational, 223, 228-229, 245
universal, 215, 219-221, 223
Granger, Richard, 305-306
grassland spiders and ESS, 162-163
group selection, 156-157, 171
Guth, Alan, 486

Haldane, J.B.S., 70, 96, 166
Hamilton, William, 166
Harris, Zellig, 218

Hart, Michael, 351-352, 399-402, 406
arguments against ETI, 400-402,
406-408

Hawk-Dove game, 158-160
Hawk-Dove-Indecisive game, 161
Hawking, Stephen, 446
Heisenberg, Werner, 450-453
Heisenberg Uncertainty Principle,
434-438, 440, 442, 548
heredity, 149
Hewish, Anthony, 3-5
hidden variables, 439-440, 457-458,
470-471, 474
Hilbert, David, 279-280
Hinton, Geoffrey, 310
Hofstadter, Douglas, 266, 300-301,
306

Horowitz, Paul, 376-377
Hoyle, Sir Fred, 5, 117-118
humanism, 329
hydrated electron, 510-511

hypercycle, 92-95

hypothetical ETI message, 543-544

Imitation Game, 261-265
incest avoidance, 176
incompleteness theorem, 279-280, 321
inconsistency theorem, 279
induction, 19
problem of, 20, 30-33
inflationary universe theory, 486-
487

information-processing machine,
254-255

Initial State Paradox, 485-487
instrumentalism, 25, 46
intelligence, 357
probability of emergence of,
357-359

internal dynamics, 256
interpretation, 272
interpretive semantics, 245
investigator interference, 128
irrationalist, 46

Jumbo (program), 301-304
Jung, Carl, 411
on alchemy, 411
“junk” DNA, 85-86, 91, 509
Just So stories, 177, 194, 196

Kalman, Rudolf, 25
Kammerer, Paul, 50
and midwife toad, 50-51
suicide, 51

Kardashev, Nikolai, 372
Kinetic Theory of Gases, 23-24
kin selection, 166-167, 171-172, 184,
195

coefficient of, 167
Kitcher, Philip, 175
knowledge of language, 231
Kochen-Specher Paradox, 552
Kohlberg’s theory of morals, 528
Kolmogorov, Andrei, 280
Kraus, John, 375

560

INDEX

Kuhn, Thomas, 39
and scientific paradigms, 39-40
and Fivefold Way, 44-45
compared with Popper and
Lakatos, 45

Lakatos, Imre, 35
and scientific research programs,
35

Lamarckian inheritance, 82
Langton, Christopher, 137
language, 211-217
common characteristics, 213
context-free, 248
context-sensitive, 248
hierarchical structure, 247-248
origins, 211-212
Rule Problem, 530
Syntax Problem, 530
System Problem, 530
language acquisition, 216-217, 495
language competence, 231
law of effect, 236
Lenat, Douglas, 306
Levins, Richard, 187, 192
Lewontin, Richard, 187, 190-192
Liar’s Paradox, 281
life, 74

functional activities, 74, 137-138
probability of, 353-357
Life game, 132-136
lifetime of civilization, 360-362
LINCOS (Lingua Cosmica), 542
linguistic determinism, 242
linguistic relativism, 242
linguistic research, 212-214
empiricists (localists), 212, 214,
231, 243

rationalists (globalists), 212, 217,
231

locality, 474

logical positivism, 27, 32
logical positivists, 31
Logic Theorist (program), 292-293
Lorenz, Konrad, 155-156
theory of animal aggression, 156,
158

Lucas, John, 320, 333
Lumsden, Charles J., 178
Lumsden-Wilson Thesis, 152-153
Lysenko, T. D., 70

McCarthy, John, 285, 327
McCarthyism, 461-462
McCracken, Daniel, 328
machine, see computer
Mahfouz, Naguib, 213
Many-Universes Theory, 485
Many- Worlds Interpretation,
453-456

Margulis, Lynn, 101
material cause, 130
Material Mode, 130
mathematical system theory, 255-256
Mayr, Ernst, 407
Meaning Circuit, 450-451
means-end analysis, 292
mechanism, 329
meme, 195
mentalism, 256
mental modules, 251
mental representations, 254-256
Merton, Robert K., 51
and norms of science, 51-52, 55
metabolism, 74, 96
micro worlds, 295-296
Milgram, Stanley, 143-145
teaching experiment, 143-145
Miller, Stanley, 70-71, 95, 99
Miller experiment, 71-73, 96-97, 100,
103, 354
diagram of, 72
Minsky, Marvin, 307-308
Mirror Hypothesis, 386
mitochondria, 101
model, 21-22, 64, 272, 338-339
of cognitive processes, 268
mathematical, 22, 24
modeling relationship, 338
Monod, Jacques, 250
Moore neighborhood, 132
morphemes, 213
Morris, Henry, 124
Morrison, Philip, 342, 387, 402

INDEX

561

multiplier effect, 197

Mystery of the Quantum World, 423

mysticism, 329

myth, 16-17

naive realism, 415, 457-458
basic tenets of, 415
naked genies, 89
NASA SETI program, 375-377
all-sky survey, 375-376
targeted search, 376
neutron star, 2, 5-6
Newell, Allen, 290, 305
Newton, Isaac, 21
and idea of mathematical model, 21
Newton’s Second Law, 417-418
Niessert, U., 94-95
nondistributive lattices, 458
normal science, 42
Norms game, 202-203
nucleic acids, 75-76, 356
nucleotides, 75, 84
bases, 75

pairing rules, 76-77, 80
nucleotide synthesis, 100-101

objectivity, 474
Ohio State SETI project, 378
Oparin, Alexander, 69-70, 96
Oparin-Haldane Hypothesis, 69. See
also Primordial Soup Theory
operant behavior, 234
operations science, 129
organelles, 101
Orgel, Leslie, 90, 95, 99, 108
origin of life, 90, 493-494
Caims-Smith Clay Theory,
109-114

creationist view, 122-124
Directed Panspermia Theory,
116-117

Dyson theory 105-107
Eigen scenario, 91-92
problems with, 93
Fox’s scenario, 98
problems with, 99

Gilbert scenario, 90-91
problems with, 93
Hoyle and Wickramasinghe
Disease Theory, 119-120
Hoyle and Wickramasinghe
Lifecloud Theory, 118-119
Oparin’s scenario, 97
problems with, 96-97, 99
Shapiro-Dyson scenario, 108
Shapiro theory, 102-104
origins science, 129
overstabilization, 361

Pagels, Heinz, 482
Pais, Abraham, 456
Panspermia Theory, 116
Papagiannis, Michael, 372
Papert, Seymour, 319
paradigm, 39-43, 64
paradigm shift, 42-44
parental investment, 164
parental manipulation, 172
Pavlov, Ivan, 233
Penzias, Arno, 476
peptide, 98
perceptron, 308-309
Petri Dish Hypothesis, 386
phenotype, 148, 150, 154
philosophy of science, 47
comparison table, 47
phonemes, 213
phrase marker, 226
phrase structure rules, 225
physicalism, 256
Piaget, Jean, 237-239
comparison with Skinner and
Chomsky, 242

and language acquisition, 241
Piagetian stages of mental
development, 239-240
pilot wave, 462
Pioneer 10 plaque, 382
Planck, Max, 419
planetary bias, 363-364
Planetary Society, 377-378
planetary systems, 346-351
double, 352

562

INDEX

planetary systems (coat. )
isolated, 350

suitable for life, 351-353
polarization, 459
polymer chains, 354-355
Ponnamperuma, Cyril, 73
Popper, Sir Karl, 32-33
vs. logical positivism, 34
population collapse catastrophe,
94-95

potentia, 451-453
poverty of the stimulus, 216
primordial soup, 92
Primordial Soup Theory, 69, 100,
127

difficulties with, 127-128
Principle of Continuity, 492-493
Principle of Mediocrity, 341, 347,
352, 366, 371, 403, 407, 411,
483

Principle of Plentitude, 343
Prisoner’s Dilemma game, 198-202
computer tournament, 200-202
private events and language, 236
probability, 365
subjective, 366

Problem of Auxiliary Hypotheses,
33

Problem of Genetic Constraints,
193-194, 196

Project Ozma, 341-342, 371
dates for, 539
Project Sentinel, 377
proof sequence, 273, 275
proteinoid, 97-98
proteins, 75-76
protein structure, 84
Proxmire, William, 378-379
pseudoscience, 57-62
hallmarks of, 57-59
pulsar, 5-6

quantum cosmology, 484-488
Quantum Description Problem,
424-428, 430

Quantum Interpretation Problem,
438-440

orthodox view, 438, 442
reactionary view, 438-439
quantum logic, 458-461
Quantum Measurement Problem,
432-434, 440, 443, 445, 453
quantum object, 440, 443
quantum potential, 462-464
Quantum Potential Interpretation,
461-465

quantum reality, 497
quantum wave function, 425-428,
440, 462

collapse of, 431, 454-455, 464,
466

quasi-species, 92

radio noise on Earth, 370
radiotelescope, 368-371
Arecibo, 379

frequency range, 369-371
search direction, 371-373
sensitivity, 371
Rapoport, Anatol, 201
rationalist, 46
rationality, 199
collective, 199
individual, 199
realism, 24, 46
reality, 417-419
consciousness-created, 449
contextual, 417
Newtonian, 417-418, 429
objective, 417
observer-created, 449
reciprocal altruism, 173
recognition physics, 419
reducing mixture, 69, 71
reification, 194
relativism, 25, 46
replication, 74
protein, 103
replicator, 148, 154
Reseller, Nicholas, 405
argument against ETI, 405-406
retarded waves, 465
Rhine, Joseph B., 467
ribosome, 77

INDEX

563

RNA, 76-77
exon, 91

messenger (mRNA), 76
replication, 89-90
self -catalytic, 90-91
transfer (tRNA), 78
Rothman, Tony, 482-483
Rucker, Rudy, 328-329
rules of inference, 269, 275
Rumelhart, David, 310

Sagan, Carl, 377-378, 382, 387, 398
and SETI program, 384
Sahlins, M., 195
Sampson, Geoffrey, 246
Sapir, Edward, 242
Sapir-Whorf Hypothesis, 242-243
Saussure, Ferdinand de, 214
Schank, Roger, 297
Schrodinger, Erwin, 424, 445
quantum description, 424—429
Schrodinger’s Cat, 445-446
Schuster, Peter, 94, 409
Schwartz, Barry, 193, 207
science, 11

ideology of, 13-14, 56
logical structure of, 13
public conceptions of, 11
as social activity, 52-56
science and religion, 62-66
comparison table, 65
differences between, 64, 124-125
possible reconciliations, 65-66
Science for the People Sociobiology
Study Group, 187-190, 197,
204, 206

scientific method, 13, 46
scientific research programs, 35-36
hard core of, 35
negative heuristic of, 35
positive heuristic of, 35
protective belt of, 35
scientific theory, 23-24
criteria for, 129
scientism, 67
Scopes trial, 121-122
Scrabble, 269-270

scripts, see frames
Searle, John, 288, 322, 334-335
selection, 149, 176
natural, 150, 154
selfish gene, 175-176
selfish RNA, 94-95
self-reference, 335
self-repair, 74

self-reproducing automaton, 131
requirements for, 131-132
semantic network, 291
semantics, 213-214
Shapere, Dudley, 44
Shapiro, Robert, 102
Shklovskii, I. S., 373, 376
short-circuit catastrophe, 94-95
SHRDLU (program), 295-296
Simon, Herbert, 246, 290, 299
Simpson, George Gaylord, 407
simulation, 338-339
of cognitive processes, 267
Skinner, B. F., 232-236
and language acquisition, 236
and pigeon guidance system, 235
Skinner box, 235
Smith, John Maynard, 158, 201
social behavior, 146
human, 146
social Darwinism, 187
Society of Mind, 307-308
sociobiology, 494-495
and animal behavior, 170
and falsification, 196
and morals, 185-186
and Prisoner’s Dilemma game,
202-203

political objections to, 186-192
and religion, 184-185
scientific objections to, 177-178,
192-198

and sexism, 182-184
Solzhenitsyn, Alexander, 206
Spectral Area Theorem, 435, 437
Spencer, Herbert, 187
Spiegelman experiment, 88-89
Spiegelman monster, 88
Spielberg, Steven, 377
spontaneous generation, 508

564

INDEX

stars, 345
binary system, 346
G-type, 346, 363, 371, 373
rate of formation of, 345-346
star- type chauvinism, 363
state, 283-284
brain, 283-284, 331
machine, 283-284, 287, 331
mental (cognitive), 283-284, 287,
331

Steady-State Theory, 483
Steiner, George, 243
stimulus-response behavior, 232
stimulus-response pattern, 254, 256
strategy, 160

evolutionary stable (ESS), 160
uninvadable, 160
Strategy of Sociobiology, 154
string, 268, 275
admissible, 269
provable, 273-274
structural linguistics, 218
Sturrock, Peter, 366-367
Summerlin, William T., 53
and patchwork mouse, 53-54
superluminal signaling, 465, 471
arguments against, 475-476
surface structure, 229
Sutherland, N. S., 328
syntax, 214, 222-223
synthetases, 88, 93, 102
system, 268
complete, 274-275, 279
consistent, 274-275, 279
external description, 254-255
formal, 268-273, 279-281, 288
internal description, 254-255
states, 256

Tarter, Jill, 375

Teilhard de Chardin, Pierre, 487
theorem, 271, 273, 275
Theory of Relativity, 418-419
General, 419
Special, 418, 463
Three-Coin Problem, 292-293
Three-Polarizer Paradox, 459-460
Tinbergen, Niko, 207

Tipler, Frank, 401-403, 487
dispute with Sagan, 402
TIT FOR TAT, 201-202
toolmaking, 357-358
Transactional Interpretation,
465-467

transcription, 77
transformational rules, 228
Trivers, Robert, 173
truth, 274
formalizable, 278
logical, 274
Tryon, Ed, 486
Turing, Alan, 264-265, 288
Turing-Church Thesis, 278, 285
Turing machine, 278-279, 285, 288
Turing Test, see Imitation Game
twenty-questions game, 416
Types I, II, and III civilizations,
372-373

UFO Hypothesis, 386
Urey, Harold, 71, 99

variation, 149

Yelikovsky, Immanuel, 7-9, 59, 61
comparison with Bell and Hewish,
9

and Worlds in Collision
controversy, 7-9
Verification Principle, 27, 32
Verschuur, Gerritt, 374
Vienna Circle, 27, 29
von Neumann, John, 39, 131,
288-290, 442-445
on thinking machines, 289-290
von Neumann probe, 401
Voyager probes videodisk, 382-385

Watchmaker Parable, 246-247, 409
waterhole frequency, 370, 376
Watson, John B., 232-233
waveform family, 425-428, 434, 437,
440

conjugate, 434

Weizenbaum, Joseph, 325-328, 332

INDEX

565

Wheeler, John A., 416, 419, 447, 482
Wheeler-Feynman absorber theory,
465

Whitley, C. H., 321

Whole Environment Evolution

Synthesizer (WEES), 86-87
Whorf, Benjamin, 241-243
Wickramasinghe, Chandra, 126
Wigner, Eugene, 445
Wigner’s Friend, 445-446
Wilensky, Robert, 319
Wilson, Edward O., 174, 187-190

Wilson, Robert, 476
Wilson’s Ladder, 175
Winograd, Terry, 295-296
Wittgenstein, Ludwig, 27-30
and logical structure of language,
27-29

and picture theory of language, 31
words, 213
WOW signal, 375

Zoo Hypothesis, 386

Grateful acknowledgment is made to the following individuals and publishers for permission to re-
produce material used in creating the figures in this book. Every effort has been made to locate the
copyright holders of material used here. Omissions brought to our attention will be corrected in fu-
ture editions.

Cambridge University Press for Figures 1.1, 7.1, 7.2, and 7.4, which are reproduced from T. Hey
and P. Walters. The Quantum Universe; Figure 2.7, which is reproduced from F. Dyson, Origins of
Life; and Figure 2.8, which is reproduced from A. Cairns-Smith, Seven Clues to the Origin of Life.

Harper & Row for Figure 1.3, which is reproduced from I. Barbour, Myths, Models, and Paradigms.

Transworld Publishers for Figure 2.3, which is reproduced from J. Gribbin, In Search of the Double
Helix.

Basil Blackwell, Ltd., for Figures 2.2, 2.5, and 2.9, which are reproduced from A. Scott, The Cre-
ation of Life, and for Figure 5.1, which is reproduced from Mindwaves, C. Blakemore and S. Green-
field, eds.

Basic Books for Figure 2.4, which is reproduced from D. Hofstadter, Metamagical Tk etnas: Questing
for the Essence of Mind and Pattern.

Reidel Publishing Co. for Figure 2.6, which is reproduced from N. Lahav, “The Synthesis of Prim-
itive ‘Living Forms’: Definitions, Goals, Strategies and Evolution Synthesizers,” Origins of Life, 16
(1985-86), 129-149.

Elsevier Science Publishing Co. for Figure 3.2, which is reproduced from D. Barash, Sociobiology
and Behavior.

W. H. Freeman and Company for Figure 3.3, which is reproduced from J. Maynard Smith, “The
Evolution of Behavior,” Scientific American, September 1978.

Harvard University Press for figure of coevolutionary circuit in the “To Dig Deeper” section for
Chapter Three, which is reproduced from C. Lumsden and E. O. Wilson, Genes, Minds, and Culture.

MIT Press for Figure 4.4, which is reproduced from B. Whorf, Language, Thought, and Reality;
Figure 4.5, which is reproduced from D. Lightfoot, The Language Lottery; and Figures 6.4 and 6.5,
which are reproduced from Communication with Extraterrestrial Intelligence, C. Sagan, ed.

Routledge and Kegan Paul, Limited, for Figure 4.6, which is reproduced from F. von Schilcher
and N. Tennant, Philosophy, Evolution and Human Nature.

Atheneum for the poem “Coiled Alitarine” from J. Hollander, The Night Mirror, 1971.

Hough ton-Mifflin Co. for Figure 5.2, which is reproduced from R. Rucker, Mind Tools, illustration
by the Design Group, Nancy Blackwell, Susan Micklem and Sarah Micklem.

Petrocelli Books, Inc., for Figure 5.3, which is reproduced from P. Jackson, Introduction to Artifi-
cial Intelligence.

Academic Press, Inc., for Figure 5.4, which is reproduced from T. Winograd, Understanding Natu-
ral Language; Figures 6.1 and 6.2, which are reproduced from S. Dole, Icarus, 13 (1970), 500-504; and
Figure 6.6, which is reproduced from G. Verschurr, Icarus, 19 (1973), 329.

Michael Arbib for Figure 5.6, which is reproduced from M. Arbib, Brains, Machines, and Mathemat-
ics, McGraw-Hill.

National Radio Astronomy Observatory for Figures 6.3 and 6.8, which are reproduced from The
Search for Extraterrestrial Intelligence, K. Kellerman and G. Seielstad, eds.

The Ohio State University Radio Observatory for Figure 6.7.

Prentice-Hall, Inc., for Figure 6.12, which is reproduced from J. Baugher, On Civilized Stars.

Windward Press, Ltd., for Figures 6.12 and 6.13, which are reproduced from Barlowe's Guide to
Extraterrestrials, W. Barlowe and I. Summers, eds.

Professor Frank Drake for the figure depicting the solution to the alien message shown in the “To
Dig Deeper” section for Chapter Six.

Doubleday & Co. for Figures 7.5, 7.7, and 7.9, which are reproduced from N. Herbert, Quantum
Reality, 1985.

John A. Wheeler for Figure 7.8, which is reproduced from J. A. Wheeler, “How Come the Quan-
tum!,” Annals of the NY Academy of Sciences, Vol. 480, 1986.

Justin Leiber for Figure 4.3, which is reproduced from his book Noam Chomsky, New York : St. Mar-
tin ’8 Press, 1975.

THE AUTHOR

John L. Casti completed a Ph.D. in Mathematics from the University
of Southern California in 1970. Following tours of duty at The RAND
Corporation and the University of Arizona, he left the USA in 1974 to
take up a post as one of the first research staff members of the Interna-
tional Insitute for Applied Systems Analysis (IIASA) in Vienna,
Austria, where he worked on problems of system modeling and applied
systems analysis. In the autumn of 1986 he joined the faculty of the
Technical University of Vienna.

His current research interests center about the development of a co-
herent theoretical framework for naturally incorporating the charac-
teristic features of living systems, self-repair and replication, into the
standard Newtonian framework generally used to model natural phe-
nomena. He currently divides his time between the United States and
Europe, where he is engaged in preparing a book on the circle of ques-
tions surrounding problems of uncertainty, randomness, prediction
and explanation in modern science.

“A DAZZLING, SPLENDIDLY WRITTEN SURVEY OF THE
| LEADING SCIENTIFIC CONTROVERSIES OF OUR TIME . . . CASTI

HAS JUMPED INTO THE RANKS OF THE NATION’S TOP
SCIENCE POPULARIZERS” MARTIN GARDNER

The origins of life. . . Extraterrestrials. . .Our genetic
destiny. . .The roots of language and learning. . .Quantum physics
and the shape of the universe. . .Artificial intelligence. . . In a
! masterful "trial by reason,” author John L. Casti presents all sides of the
most important and vital scientific debates raging in the world
today — scrutinizing six perplexing “great questions” in the most
engaging, astonishing and accessible amalgam of science
and literature since A Brief History ofTime.

“A DEER CAREFUL AND PLEASANT CONSIDERATION OF WHAT
SCIENCE IS AND HOW IT IS DONE. IT WOULD MAKE ANYONE WANTTO
BE A SCIENTIST." Isaac Asimov

“EXTRAORDINARY... BROUGHT OFF WITH CONSIDERABLE ERUDITION
AND WIT... ABSORBING READING" Kirkus Reviews

I “ASTOUNDING IN THE BREADTH AND DEPTH OF ITS PANORAMIC VIEW
OF SCIENCE" George Leitmann, University of California at Berkeley

\ “A REMARKABLE EVENT... A CONCISE AND ELEGANT ACCOUNT... OF ;
IMMENSE VALUE" Benjamin Rode, San Antonio Star

Internet Archive Audio

Featured

Top

Images

Featured

Top

Software

Featured

Top

Texts

Featured

Top

Video

Featured

Top

Mobile Apps

Browser Extensions

Archive-It Subscription

Save Page Now

Full text of "Paradigms Lost. Tackling the Unanswered Mysteries of Modern Science"

See other formats